Adamw

https://www.aim-sportline.com/download/ecu/racing/LinkG4_101_eng.pdf

Suspect your max clamp of 30% isnt enough power to fully open the throttle.  Can you set that to 90% and give it another try.  

Yeah, if it is <1ohm it is definitely a dumb coil.  The reason for the no start would be with no ignitor the spark would be very weak.  obviously just strong enough to ignite the high volatile starter fluid sometimes but not e85.  
The ignition system change going forward is proobably personal preference.  2 x decent 4 channel ignitors end up costing about the same as completely new smart coils in my experience so cost is a moot point.  Stock coils with ignitors will probably be the neatest and require less wiring changes.  The potential downside maybe the stock coils have less energy than the well proven LS coils (I have no experience with the Ford coils and would only be a consideration if high boost was intended in future).  The LS coils will need a some more wires added and brackets and HT leads made.    

If IAT is not the cause of your issue then the other common cause of lean hot restart is due to the injectors getting hot since they are sitting in a very hot environment and attached to a hot cylinder head, with very hot intake valves right underneath them and without the usual fuel flowing through them to keep them cool. 
The way to work around this is to use a 3D IAT trim table with IAT Vs engine runtime on the axes.  Then at hot IAT's you can add say an extra 15% fuel at initial startup and fade it out over say 2 mins.  

Yep, pretty much all Fords are dumb coils.

If the Boost source is set to MAP then the target will be MAP.  This is the most common option as MAP is what determines the torque (cylinder pressure) and also the factor that you need to consider against engine or driveline component strength. 
 

I'll call in @Vaughan to see if he can take a look at the reason for no dwell.  Im not in my office today so dont have all the info or gear I need to test what is going on.  On first peak I thought the scope looks like it should be acceptable to me.  Possibly trig 1 arming threshold is a bit too high, but lets see if Vaughan can test on the bench or has any other ideas.  Otherwise I can take a closer look at the end of the day. 

Ive had this on two K engines now.  The first one I discussed above was in a civic circuit car, refitting the factory balancer fixed that one.   The second one was a Juno sports car so no almost no flywheel and no room for a damper on the front, on that one the owner had to mess around trying a few different ideas to isolate the throttle.  The final solution was a bracket to support the manifold (it was underneath and connected to the block with a rubber mount at one end), and he also made some sort of soft plastic/urethane or similar spacer between the throttle and manifold. From memory it also had some soft "washers" under the nuts a bit like a cam cover has.   

If IAT is not the cause of your issue then the other common cause of lean hot restart is due to the injectors getting hot since they are sitting in a very hot environment and attached to a hot cylinder head, with very hot intake valves right underneath them and without the usual fuel flowing through them to keep them cool. 
The way to work around this is to use a 3D IAT trim table with IAT Vs engine runtime on the axes.  Then at hot IAT's you can add say an extra 15% fuel at initial startup and fade it out over say 2 mins.  

I can possibly help with the flat shift and shift light if you attach the tune, a log, and some info about the car, what engine and what ecu.  
The flames I dont have much interest in.  

The Haltech gauge is locked to the Haltech protocol, it will need a few custom streams set up to duplicate the haltech messages for the channels he wants.  
 
Can you set up your CAN mode tab like below then go to the CAN devices tab and see if it finds a Lambda.  Suspect it maybe programmed as something other than Lambda 1.  

 
 

Yes the pins are fine.

The value generated by the table is also linearly interpolated between cells, so it generally doesnt matter where you are working.  If there is some area of operation where linear interpolation between two cells doesnt give the outcome you desire then just add a new row/column at the value you want.  

Load your map back in to see if it’s still working 

Alex, instead of setting the Analog input up as "Voltage 0-5V" as I did in the example above, you could set it up as "GP Input", label it "Launch RPM" and assign a Cal table.  Your cal table will be Voltage vs RPM.  You can then set up your LC table with "Launch RPM" on one axis (instead of voltage).  Your analog input should now be logged as RPM instead of voltage.  I can post some screenshots later if you cant work it out from these instructions.

Wheelspeed input via CAN is fine.  Most of the plug-in ECU's for modern cars get wheelspeed via the ABS ecu.  Not too many wheelspeed to CAN devices in the aftermarket though.  ECUmaster and AEM have one I think.  
CAN DI's can be used for virtual auxes.  
Since there are so many different ways to acheive things, the best way to see if something is possible is to try to set it up in the software yourself, you will quickly find what is possible and not.  List below is what inputs can be received over CAN.  You will notice Analog input and digital inputs options are pretty limited compared to hard wired inputs.  

Can you attach your tune and give us a log of it happening.  Start the log with engine running, turn off ign (leave log running), wait a few seconds, turn ign back on, (hit F3 to connect if not in auto connect mode), then start engine and record for a little while longer before stopping log and saving.  

The wire from the ignitor is not a "ground", with the stock coil this wire would only be grounded during the dwell period (ie, about 3ms every TDC).  Smart coils should be grounded to the cyl head.  

In that case your setup will be working correctly as is and needs no changes.  Except I'm not sure how you assigned an analog input to lambda 5.  
FYI, the exhaust pressure compensation is done inside the CAN lambda devices.  The ECU sends calibrated pressure to the CAN lambda, they take it into account in their O2 calculation and send compensated Lambda back to the ECU.  The AFR500 does not have any of this functionality so will always assume exhaust pressure is at sea level atmoshpere.

The “find devices” tool is only for Link devices and CAN keypads.  
 
The VDO gauges look like they are expecting J1939 or CANOpen messages.  The G4X will likely be able to generate some of these messages but they are large complicated standards and I don’t have a lot of exposure to either.  J1939 is often used in agricultural machinery and CANOpen is mostly used for industrial control systems.  
Can you share a link to the gauge software and I will see if it gives us any clues to help

Yep looks correct.  I would just change the error high to 5.00 as I suspect during overrun fuel cut etc it will go above 4V.  

You have got Lambda 1 calibration set to Linear Cal 1, but Linear cal 1 is not set up - it has no units and zeros for the output values.

That would be correct assuming the coils you call 1/2/3 are connected to the cylinders in this order:
Coil 1 = Cyl 1 & 6 Coil 2 = Cyl 2 & 5 Coil 3 = Cyl 3 & 4

I would try an aux rather than an injector drive in that case.  They are only rated at 2.2A continuous, but since you wont need 100% DC it will likely be enough.  The reason for aux rather than injector drive is the auxes have a flywheeling diode connected so the solenoid will respond faster.
 
No you will need to connect +12V to one side of the solenoid and the ecu drives the other side.
 
It is correct.  Aux 1 is used for the two speed fuel pump on GTR's and in R33 RB25's it is used for the VTC.  

Yep, will stay in sequential if cam sensor is lost.  It will even start up in sequential with no cam sensor (but will only start approx 50% of attempts).  
Although any tooth could be removed, I would suggest the tooth that is under the sensor when the crank is at about 90BTDC or 90ATDC will be the best option.