Adamw

A NOS solenoid would typicaly be done using a GP PWM.  A basic example below.  The solenoid is active when TPS is above 80% and MAP is above 200Kpa and there are no limiters active.  The PWM table you can set up however you like.  Use a 4D fuel table to add extra fuel when the NOS is active.  

Some inspiration for @Arron Eades .  The preferred ECU mounting system for every Le mans car since 1990 :

For the solo2DL set the ECU stream in RS3 to Link CAN BUS BASE.    
The CAN bus should be a trunk of twisted pair with any stub off the main trunk kept short. 
You can do it like this:

 
Or I often find a "daisy chain" type variation easier so there is effectively no stub at all, something like this:
 

It will work.  MGP would be the recommended option though. 

It appears there is an issue with your TPS main signal.  If you have already replaced the throttle then it would have to be a bad connection to the TPS main sensor somewhere.  The throttle position control uses the TPS main sensor, the TPS sub is only used as a comparison for error checking.  In both of your logs the TPS main gets erratic just before the engine stalls, but if we look at TPS sub it is not erratic and is actually sealed shut at zero.  So it looks like TPS sub is actually the correct one and TPS main is wrong.  
I think that could also explain the MAP signal taking a long time to bleed away the vacuum yesterday as with the throttle sealed shut there would be vacuum produced until the engine completely stopped.  The engine is still actually turning after the log shows zero RPM, this is just the point where the trigger signal is no longer reaching the threshold.   
Since the TPS sub signal looks reasonably realistic I think that suggests the 5V and gnd is probably ok, the issue is more likely with the TPS main signal connection so that is what I would concentrate on first.  I would start by setting up a big gauge on a PC Link screen with TPS main displayed, watch that while yanking/bending/stretching/shaking the loom and connectors around the throttle, ecu, and the path in between.  Hopefully you will see it go crazy when doing that.  
As Vaughan said yesterday this is just idle ignition control hitting the max clamp trying to keep the engine alive.  
 

Ah yes, sorry for the bad info.  Im so used to not having that reallocation functionality in our older software and have never needed it so forget it is even there.

With G4X ECUs on the latest firmware (comes with the G5 PCLink install) you can reallocate which ignition and injection drive is used for what so if you can use any of the Ign 5-8 pins on the expansion header as your additional ignition outputs as long as you reallocate them correctly in the Ignition Reallocation table.
Example below uses Ignition 1 output for cyl 1 coil, Ignition 2 output for cylinder 2 coil, ignition 5 output for cyl 3 coil and ignition 6 output for cyl 4 coil.

Yeah something very wrong here.  TPS delta would usually be less than 1%/100ms with the engine running steady state.  Your MAP delta doesnt look too out of the ordinary for an ITB engine, so I would say that is probably ok and wont cause you any tuning issues.   
A couple of thoughts for the TPS:
Unplug the MAP sensor and unplug/completely disconnect the lambda controller (including its 12V & gnd) and run the engine like that with just the TPS connected.  See if TPS noise improves with those two taken out of the equation.  My thought here is possibly if the MAP was wired wrong you could have MAP noise superimposed on the TPS signal - or if it is some "fake" sensor it could be doing anything to the 5V or Gnd.  I've had a fake denso coil sending 400V back into the ecu output every time there was a spark before so I wouldnt rule out anything...  The reason for disconnecting the lambda controller is the Innovate lambda controllers are a complete POS - there is no isolation between the high current PWM heater control and the analog output so the ground connection can add noise to all analog sensors if grounded to a point that has a path back to sensor ground.   Backlash in TPS coupling - Check the TPS is a tight fit on the "D" shaft, this is especially important if the TPS is one that doesnt have the internal spring to soak up the lash.  If there is any doubt then put some RTV in the D before assembly and let it dry before running again.  I have "fixed" a couple of Nissan sensors this way that had noise due to a loose fit on the shaft.  Vibration in the cable/linkage system - Im not familiar with the RHD kit, although from what I have seen Rama does make some nice throttles.  But assuming this is a 911 we are talking about then I imagine the linkage is a complicated contraption and I vaguely remember they have that horrible cable or rod down the the centre of the car that I think is pulling from a chassis mounted bracket instead of an engine mounted bracket like a more conventional cable system?  So potentially some vibration could be introduced into the throttle shafts via the linkages or any movement between engine and chassis?  Can you feel any vibration in the throttle shafts with the engine idling?  I have fitted a "bolt-on" Jenvey kit on a 911 before and I had to cut up most of the linkage, shift arms around, extend levers etc just to get full and equal movement on both sides so I know the linkage can be a bit of a challenge on these. Alternator - disconnect the alternator and give it a short run, possibly AC from a bad diode or something could cause some noise in the electrical system somehow.      The TPS itself - I have had trouble with noise from a generic-looking BMW sensor before - although it was a used sensor of unknown miles, and I have also seen large voltage drift with temperature and regular failures from the cheap looking vishay/colvern ones that many aftermarket throttles use.  So I always use the contactless ones on ITB race engines now.  I think in this case it is unlikely a sensor issue since you have tried 2 already, but if there is any doubt about either of the ones you have tried and all of the above suggestions dont lead anywhere, then I would try swapping to a variohm contactless one to rule that out.

The idea of a seperate dog unload and main shift stage is with dog engagement gears you want an short/sharp torque reversal just to get the dogs separated, once they have released the goal is just to slow the engine down enough so the shaft speeds arent vastly different for engagement.  
In a synchro gearbox there is no undercut on the dogs and the clutch does the unloading so in theory you dont need a different cut for the disengagement Vs engagement.  I would set the dog unload duration to 0, so you dont need to worry about any of the dog unload settings, your whole shift will be handled by the "main shift" settings. 
For a synchro flat shift, the start shift mode would usually be "digital input" using a clutch switch, the end mode can be digital input or time.  Digital input means the cut will be active for the whole time the clutch is pressed, time means you would have a table controlling how long the cut is applied for for each gear.  Both have pros and cons - to get the shortest possible shift in digital end mode the clutch switch needs to trigger in the right points (adjustable) and that will change with clutch wear etc.  For the time based end mode the driver needs to be fairly consistent with how fast he pulls the lever and the timing between pedal and lever movement. 
Start with about 80% severity and 200ms duration if using time mode.    

That suggests PC Link isnt aware of some settings that are in the firmware.  Can you tell me what firmware and PC Link version you are using.  You can find these by going to >help>ecu information (need to have a map open from the ecu or be connected to the ecu) and >help>about.  
 
The SI Drive mode is available on CAN AN 1.  Notes below to show how it is enumurated.  
There are many different ways to utilise this for adjusting boost control such as switching between 3 different boost tables, interpolating 2 boost tables, or just referencing CAN AN1 on a single boost table.   And any of these option could use closed loop or open loop.  An example below using a single table varying boost target with CAN AN 1.

Im pretty sure this trigger mode uses a count type sync so I suspect either rising or falling edge would work for trig 2 in terms of ability to sync and start.  The ecu just checks whether it saw 1 or 2 trig 2 edges in the last crank rotation (between the missing teeth).  Swapping from rising to falling would effectively inverse the count result so the offset should change by 360 deg.
I dont think the 2GR mode does a wasted spark start so if it is running, it is running direct spark and the offset is correct.  
For the DI's, It sounds like possibly the voltage level is not crossing one of the thresholds.  You can check what voltage is reaching the ecu by backprobing with a multimeter on the DI pin, and the engine not running, just pull a sensor out and place something iron in front of it to get both the high and low level.  From memory with G4+ it needs to go above about 2V and below about 1.2V.  
 

Yes you can repurpose/extend the factory IAT wiring.  The X series pressure gauge has both a CAN bus output and an analog output , so you can wire it to the CAN bus or and analog input on the expansion loom.  

Yeah, no point in connecting the boost gauge.  
For the other two, just connect AEMNet + to CAN H and AEMNet- to CAN L.  

Yes, most likely a sensor issue.  The only time I have seen error 54 is with a damaged/old or fake sensor.  

It's like extracting blood from a stone sometimes...
What version/year/model WRX? Does it have traction control and DSC?  And is it a plug-in ecu or wire-in?

I moved to the G4+ forum as you posted in the G4 forum and there is no G4 Fury.  So I assume you have a G4+ or G4X.
The Fury has aux outputs capable of both high side drive (+12V) and lowside drive (switch to ground).  Aux 5,6,7,8 can do high or low side, all the others can do lowside only.  Obviously these dont power the fuel pump directly, they just switch the fuel pump relay on and off.  
 
Lowside drive wired like below is the most common:

What function/sensor are you talking about?  
 
If you are running the stock controller you can leave it as is.  

Attach a copy of your tune, a PC log of it cranking, and a trigger scope captured during cranking.

The Fury has a builtin lambda controller, it doesnt need a CAN lambda controller, you can wire the LSU4.9 sensor directly to the relevant pins on the ecu.  

Fuel pressure sensor is connected to an aux?  
The AC pressure switch is usually fitted between the AC request switch and the ecu, so idle up isn't activated if pressure is lost and you get proper clutch delay etc when cycled on the high pressure side.
Some other stuff could be simplified such as clutch switch could just be connected to ground.     

To make it easy to get started on the forum each user gets a small storage allowance to allow them to attach log files, tune files and pictures to their post.  When you have used up that allowance and try to attach a file you will get an error message stating your file is too large to upload.  The way forward is to upload the file to an online drive service and use a share link to the file.  Google Drive and Onedrive are probably the most common free options in western countries so I have done a quick guide on those two here.  
There are many other options such as Dropbox, iCloud, WeTransfer, etc that usually follow a similar process.
How to share a file using Google Drive and Onedrive.mp4

Yeah something is a bit funny with the multiple conversions going on I think.  It should only be out by a max of about +/-2°F difference, not 13.  If you temporarily set your units to metric when you initially set up the table axis it should give you something closer to expected.  
As an example, to acheive the axis values that exhibited the problem for you 160/170/190/200F, I set up the axis in °C to 71/77/88/94.
  
 
Which gives this after setting units back to US:

I have done a quick test and the keypad counts are received correctly with the 3 separate messages/ID's as you have proposed.  

Yeah its not going to work wired like that, you've created a path to back feed.  I've attached a video to try to explain better.  
 
Backfeed 2.mp4

Yeah I think it will need to be powered up whenever the ECU is live.  Part of the CAN standards is an error checking system, if no device acknowledges the messages the ecu is sending it must report a transmit error, once it reaches 255 transmit errors then it goes to "bus off" state.  The bus wont be reset until it is power cycled or you open the CAN screen and hit apply again.