SchuKingR

That is not a "fix". Traditional mode is completely different than modelled mode.  In modelled mode the numbers in the fuel table number represent % volumetric efficiency.  This is how much air the engine is flowing, not how much fuel.  It is not possible for a for a typical road car engine to have a VE higher than 150%.
If you are using E100, then most of your problem is you have the stoichometric ratio and fuel density set to gasoline values.  The other issue is your ethanol sensor is using the wrong edge so the fuel temp is exagerated.

Test sensor by applying compressed air to it if possible, or double check with a mechanical gauge if feasible. 
If the engine was not primed with oil before installation it can take several minutes of cranking to build oil pressure as @David Ferguson says with the spark plugs all removed.  Alternatively, reverse prime the oil from the sensor location via a drill pump or similar device.  Personally I made up a rig using an oem oil pump for a 5SFE mounted to an aluminum plate I can spin with a drill and suck from a fresh quart oil container and into a spin on adapter in place of the oil filter.

https://www.instagram.com/p/CtzLTTExtkk/?img_index=1

had the error at longer drives again. Changed to a contactless sensor then. Works flawless until now.

If you need any technical documentation on the G5 Voodoo Pro, you can find it here:
Technical Specs - https://linkecu.com/g5/voodoo-pro-specs/
Quick Start Guide - https://linkecu.com/wp-content/uploads/2023/08/G5-Voodoo-Pro-QS.pdf
Installer I/O Table - https://linkecu.com/wp-content/uploads/2023/08/G5-Installer-IO-Table.pdf
LED Codes - https://linkecu.com/documentation/Voodoo Pro LED Codes.pdf
ECU CAD Files - https://linkecu.com/wp-content/uploads/2023/08/G5-Voodoo-CAD-files.zip
Dimensions Image - https://linkecu.com/wp-content/uploads/2023/08/G5-Voodoo-Pro-Dimensions.pdf
 

With too much gain the throttle will move faster than the engine responds to the change in air flow so you end up with over-correction which causes an oscillation.  The antistall gain kicks in if the RPM drops 150rpm below the idle target.  Your idle gains are quite high compared to what is typical for E-throttle, although I suspect your odd setup with a linkage and small butterfly in the mix probably means the usual rules dont apply.  Typical proportional gain is 1-2 and about 2-3 for the anti-stall.  I have never really noticed the anti-stall to have a lot of affect.   

I would change your anti-stall gain to a much lower value - currently all 25, maybe try 5, and also raise your overrun fuel cut deactivation rpm table values by like 400rpm warm at least.  Your cold stalling shows the rpm spike from 450rpm to 4000+rpm - not sure if that actually happened or it was a trigger mistake.

^That is correct, the FD's secondary throttle plates are mechanically linked to the primary throttle plate, but they should not start to open until the car is revving well past a high idle speed on a cold starts (about >1500 RPMs).

I agree with KO that it looks like it needs at least another 0.5% in the top row of your e-throttle target table. I would make the idle ign a bit more aggressive too, something like below.  But there are a couple of odd stalls in that log which make me think there may be something else going on sometimes, for example at time 16:21 it has been idling with the car stationary for some time, with throttle at about 4% and ign timing at about 4deg, then RPM suddenly starts to drop, ign timing ramps up to 13deg and the throttle opens to 16%, but it still stalls.  Is it maybe not enough accel enrichment or similar?  You could possibly do with just a tiny bit more integral gain on the e-throttle settings too, maybe try 0.07, you can see a little long-term position error in some places where TP is sitting at about 0.5% lower than target.  DBW with linkages is sometimes hard to eliminate small error like that due to a little bit of backlash in the linkages. 
Only the primary port butterfly is working at idle right? 
 

When it's idling at the target rpm, the throttle is at 3.9% open but your idle base position is 1.6 with a E-throttle target of 1.5 (total to 3.1%).  Normally I would say you should target a higher value than what it actually requires to idle correctly - like set the minimum E-throttle target to 3.5 and idle base position to 0.8-1.0 so that you get a total of 4.3-4.5 and then the close loop idle will bring it down slightly as will the timing control. Your e-throttle target table looks like it's set up to limit throttle position based on time at WOT, but you can gain RPM targetting to help as well by going to 2 tables and have them switched by your virtual aux 1.  Your base timing table is all 3 degrees around idle.  I don't know if this is normal as I've never fully tuned a rotary engine, but I would increase all the base timing below idle target on the base table to improve it's tendancy to idle up if it drops into the lower rpms before idle control has a chance to improve things.  I'm attaching a link to your tune file modified in these ways you can take a look at or try.
https://www.dropbox.com/s/wkktx1hxwpwk1dc/FD3S_Thunder_MAP_V54-Modified E-Throttle Target.pclr?dl=0

You can usually tell just from the lambda trend if it is an "amount of enrichment" Vs "length of enrichment".  I only have a couple of old logs from our FD on my laptop that arent real good examples but may help explain...
In the time plot below you can see the secondary injectors kick in where I have put the pink arrow (sec PW jumps up from zero).  For a short time after that you can see the primary PW sits above the sec PW for the length of the yellow arrow before it steps down to match the sec PW - this yellow arrow area is the staged accel enrichment.  In Staged injection settings, the "accel fuel" affects how high that step in pri PW is and the "Accel time" affects the length of the yellow arrow.  
You can see in this example the lambda dips rich right around where the accel fuel is.  

 
Now if we zoom in on that same time plot to have a closer look at this area...  Where I have placed the yellow cursor line on the plot you can see the lambda is bang on target, and also to the left of that cursor the lambda is reasonably close to target.  To the right of the cursor the lambda dips rich for a while and doesnt recover until a little while after the accel fuel finishes.  So my interpretation of this plot is I had something close to the right amount of extra fuel initially as the lambda stayed close to target for a while after secondaries where enabled ("accel fuel" setting is about right).  But that accel fuel is being added for way too long - we could have removed that extra fuel where the curosr is - or even a little earlier. So the setting "accel time" is too long - by proportions I would say the accel time wants to be about 1/4 to 1/3 of whatever it was set to when I done this log.

Adam,
Related to the OP's questions, is there a recommended approach for adjusting the staged injection Accel Fuel & Accel time parameters? I too have been struggling to get my FD to stage the injectors in as smoothly as I'd like - mine typically trend way too rich at the initial staging point, which I figure means I should be reducing the accel fuel and/or time, but is there a preferred approach as to which parameter to tweak first? What should I be looking for in the logs to inform me which change (in accel fuel or time) would be more appropriate? I don't have the luxury of a dyno to work on, so I'm trying to make my road tuning iterations as productive as possible.
Thanks,
Pete

workaround is to use a CAN adapter I think. works really well
 

I know it's G4+, so no real updates incoming in the future. But can you add that to the list for a possible future update of PCLink?

A buddy of mine uses it with a G4+ Fury in his RX7. He is quite happy with it. It's also possible to do CAN switching with it. Changing boost maps for example.

You have a plug-in ECU that is designed to work with the factory wiring and factory assigned pin functions.  There are two expansion ports to give more IO and on the GTR generally there are several other spare IO are available such as the MAF and O2 inputs.  We state the expansion IO that is available on the webpage for each specific plug-in ECU, and the plug-in ECU's are priced differently depending  on how much IO they have.  Ie a VR4 ECU is much cheaper than the WRX107 due to the IO available.  The ECU comparison chart here: https://linkecu.com/products/ecu-comparison-chart/ also states on the plug-ins not all IO is exposed.  
The reason we cant have end users messing with PCB's is the PCB design can have a massive influence on the the performance and functionality of the ECU and our reputation has been built on getting that right and that only happens with many years experience.  If for example you took your car to a tuner with your home designed PCB and the tuner found it had 20 deg of ignition scatter (due to say a DI track running beside a trigger track) - the first thing he is going to say is something like "This Link ECU has shit ignition control", it is Link that will get the blame and the reputation wiped out, not the inexperienced PCB designer or invalidated adapter board...  
If you need more IO you can use CAN bus or can return it to us for modification.  
No ecu manufacturer on the planet will give you internal schematics.  
 

I've love this change to be back-ported to the G4+ too, please!

Thanks

(These Piston motor things are just confusing sometimes....Next cars a Rotary!)

Many assume because there is a spark every crank rev that the otto cycle of the wankel engine is 360 deg - often it is thought of like a two stroke engine.  However many forget the Wankel has 3 combustion chambers and the rotor revolves at 1/3 of crank speed so it actually 1080 deg for a full otto cycle.
When you inject fuel into a fresh chamber on the intake stroke, it takes nearly a further 2 crank revs before that charge reaches the other side of the engine compressed at TDC ready for spark. 
So you cant just cut fuel and spark at the same time - There is fuel and air already sitting in the chamber from 2 revolutions ago...
The simplified sequence:  The rotary limit mode starts by cutting fuel first, but continues to spark for the next two events since there is already fuel in those chambers, then it cuts spark.  When RPM has dropped below the limit and we want to restore power, then it enables fuel first, then waits 2 spark events before enabling spark again.  
This strategy is the same as the Mazda factory ecu's use, it is the safest for stock engines as very low risk of backfires or lean combustion events, but it is also relatively "slow" and granular so not commonly used for motorsport type applications where the seals have been replaced with something more robust and can handle a more conventional limiter strategy.  

Yes stoich mode dithers 0.02 lambda either side of target.  It maybe a bit slower than auto mode, but generally works well.

I will add a little more info here since I have quite a bit of experience in this strategy.
I agree with ECU settings as per Scott's suggestions, but I would say nearly always it would be worth while to have a "MAP sensor" connected also.  In this case however you don't connect the MAP sensor to manifold, you connect it to the airbox (your FPR reference should also be tee'd to this same measurement point).  The ecu and FPR will then keep fuelling more consistent when the airbox is subjected to aerodynamic effects.  If airbox and general intake design is good I often see 4-10Kpa positive pressure (ie you will need 4-10% more fuel) on high speed sections so it can be a significant effect.

Thanks a lot for the reply. I do understand all the other variables and for some there are ways to process the data in order to eliminate some types of errors.
However i am more interested finding out the actual calculation taking place for these 2 parameters so that a more educated choice can be made between those two. Maybe one of the software engineers can give some input here.
From what you show however (which is if i understood correctly an engine that is VE (speed-density) tuned but still happen to have the MAF sensor installed but it's not used in the load calculation) is that the "Air per Cyl Estimated" to be the closest to the realistic flowrate in my opinion, due to the very close similarity of the curve. If there was a systematic error present it would show here as an offset, but if there was a characteristic error the curves would show noticeable differences.
Which brings me to the point that i should generate the calculations for some period of time because only checking at one point is extremely likely to show big errors like the 10% i calculated. I knew about this, but i figured to ask first and just give it as an example.
Lastly, i've asked in the past about this but it would be nice if we could setup custom parameters with algebraic functions in the PClink software so these calculations can be done without having to export the data csv and do it in excel. Also you should allow us to make a custom gauge graph where we can put any parameter on the axes with a stretchable underlay image so that the compressor curve can be viewed in realtime. I know you put functions in G4x but i think these are functions for the ECU, this is something that can be done on the PC software side and should be easy to put as an update for both PClink versions. Please:)  

Would be great if Link made a wifi/bluetooth unit that connected to the normal Can programming port and allowed wireless programming. That would be awesome.

I see G4x already has this feature, but it'd be great to add it to G4+, and I don't think it'd take much (if any) CPU resource to make it happen (unlike math channels).  Basically, if we could use the duty cycle of a GP PWM output of any aux output, then it'd allow for more complex logic when deciding boost, or accelerator position, or traction control slip target, or ...
As a corollary, it'd be nice if Virtual Aux had GP PWM capability too so the output of those tables could be used as well without having to dedicate an Aux output to this purpose.

For basic stuff, I use a free online one called scheme-it, from Digikey.  You just need to set up a log-in.  One of my drawings as an example below.  

To jump in on this I've always felt the advice and guidance from the team (Adam) to be 2nd to none, I've always had the help and resolutions I've needed and would in turn recommend link to anyone.