To be fair, I don't think most on that side were relying on dyno results. It's mostly relying on REST API numbers being lower than 691 hp. I don't believe there were any really conclusive dynos of the P85D (all of them were lower or the same as what the P85 got before at ~430whp, which isn't that believable).
I have suggested that if someone can do a dyno while pulling REST numbers at same time, that would be very telling. The REST should hit the 550hp mark (which there is agreement is the peak mark for REST) and then that can be matched to the dyno result, which would pretty much tell the peak at the wheels.
Like VW, the MS has special software that detects when it is on a dyno. The car has a special bottle of noxious gas that it releases during the pull while simultaneously reducing output to match forum expectations.
Maybe it just proves coming up with one hp number using traditional methods isn't such an easy thing. Further proof that an actual SAE standard is needed for EVs. Something Tesla didn't have at P85D launch and still doesn't.
These cars are dangerous on the dyno with regen. It's like hitting the brakes, which you never want to do since it makes the car try and launch off the dyno. You shouldn't take your foot completely of the accelerator at the end of the run. It needs just enough pedal to stay out of regen.