Lambda target of 0.78 with 5-7 degrees of timing at 1.5-1.7 bar boost is not how I would do it.
All I meant was that when I tune I set a reasonable target, and approach that target from the rich side to avoid a melt-down.
Your first pull was without large issues.. see where I put the marker, close to 900 degrees EGT and only 5 degrees timing, increase the timing and EGT should drop. Also, if the tuning reveals that you initially hit lambda target very well, and then you're suddenly getting leaner - then it's time to stop. It could be mis-fires, knock, a bad (overheated?) coil, or something else that needs to be adressed. 4.0 ms dwell, is that too much for these coils ?
The MAP signal varies by 20 kPa, this is with old firmware with no filtering. Still too much and indicates a mechanical problem, the wastegate valve might have high friction in the valve guide or simple be badly matched to the flows and pressures in this system.

The following pulls were not nice.. the lambda target is even richer than 0.78, at which point regular gasoline will not combust properly and you will have a slow burn, a low ignition advance makes even less sense at that point. Torque peak might be at 3000-4000 rpm somewhere, so it's a point of caution but maybe not this much.

I would upgrade to 1.1.81 or .84 (just released for testing in VemsTune, looks very promising), use less dwell and boost and see if you get a more stable working engine.