I am getting AWD for the performance. Unlike with the P100D, right now it is interesting that the non P AWD Three has the P hardware.
Yes, the two Three runlog samples are the same. We have a third sample since updating those metrics. It looks the same. I had hoped the API runlog disabling was a bug, but the longer time it goes disabled, the more likely something is up. Still, I doubt that the "feelings" that the torque increases are accurate. I find it unlikely that Tesla would have a mechanism other than fixed torque and max power.
The rollout, combined with the quarter second sampling on the API makes for some quarter to half second variation on the 0-60 time, but yes it is faster than 5.1. I really only use these (or anyone's) metrics to look at torque and max power. Hopefully someone cracks the Three CANBus. It would be interesting to compare torque hold and launch timing at hundred second increments between PS and P3.