The recent rear-ending accident video I uploaded generated a lot of questions that could not be answered adequately with limited logs available at the time.
Had to resort to some dark sorcery to resurrect the complete logs and got help from @pEEf wrt decoding.
It's controversial

The people that bet on driver asleep were probably correct. The driver did NOT hold the steering wheel and the autopilot did a "strikeout", BUT when that happens the car does not stop steering (unsafe) and continues in lane.

e.g. example here:

What made matters even worse - the driver had their foot on the accelerator pedal gradually increasing the pressure and increasing speed. While normally AP does not operate above 90mph, but it will keep steering you don't take control as in the video above).

With all this in mind here's the revised video with canbus data overlaid (the red flashing is periodic "autopilot fault" messages). Driver is forcing the speed to remain high, the autopilot does steering while beeping like crazy.

I am sure @Tesla already knows, but this is snapshot id 7c47b0f1-c0bf-4835-bbbd-58133ae311e0-0000
and if they want to add any additional details or corrections, they are more than welcome.

The Honda was also found meanwhile and it does not look pretty.

The biggest gap in this whole thing is - the car clearly sees an obstacle, sure it sits in the same lane, but why not override driver accelerator input pedal misapplication-prevention-style and just match speed? Sounds like a safer decision? Hopefully this could be fixed?

And finally with all this extra knowledge and material I now have, I just had to revisit the "night rocket" case too.
It also was operated on NoA with "driver pedal override" active" to circumvent speed limit enforcement.

https://twitter.com/greentheonly/status/1261456934840008719

oh, btw forgot to ask - how is this type of accident reflected in @Tesla safety statistic?

Is this considered "on autopilot" or not?