Fun but there are a few things that are really wrong with that graph.
First, if we're to believe that peak hp was at the point you let off the throttle, then you should have run it up to a higher speed because in theory you probably let off before hitting peak.
That said, peak power on a P85D with a 90% SOC occurs right at 44 MPH when the battery is pulling 414KW for a brief moment before starting a very gradual taper that is almost flat but not quite. It climbs very rapidly to 402 by 33 MPH, then gradually rises to 414 by 44 MPH and by 74 MPH is back down to 409(very gradual taper).
This is REST api matches this graph exactly produced by an instrumented VBOX run with a roof top GPS antenna:
This run was done at 65% SOC. I haven't had a chance yet to do them at higher SOCs because I was on a long 4th of july trip and then dropped the car off for service immediately after that. Peak hp is 455 at 44 mph matching exactly when it occurs from the REST data as well.
Also, what weight did you use? I'm using the published curb weight 4630 + 200 (my self). This could be off. However even if it's off by 30 lbs, it would almost have no effect on the result. Still, I plan to get it weighed at a racing shop in the next few weeks to really make sure. The thing that makes it more uncertain with an ICE car is that the published curb weights are dry, so you need to weigh the car with fluids and gas in order to be accurate. The Tesla, though, won't change. However, they may only provide curb weight for the P85D with a specific option set and things like pano roof or no pano roof will make a small difference as will the presence or absence of the subwoofer of the premium sound system. So it's still going to be necessary to weigh the P85D if we want to get as close as possible but even if we don't, we'll be with 1% or so.
A few other things. The REST API's KW reading vs what was measured to the ground shows a loss of a mere 10%. Normally, drivetrain loss is calculated from the motor shaft to the wheels. A rear wheel drive car is typically about 15% if you dyno in the 1:1 gear which is typically 4th gear. An all wheel drive car is typically 20 to 25% given that it has to go through many more drivetrain components such as transfer case, an extra differential, 1 extra drive shaft, and two extra half shafts.
So seeing only 10% total loss from the battery through the DC-DC converter, through the motors, through the reduction gear, through the open diffs, and finally the half shafts is truly remarkable but not totally unexpected since this is still a lot less stuff to get through than any typical AWD drive.
- - - Updated - - -
If this holds consistent, then when I repeat it again in 10% increments it should show 503 hp estimated at the wheels with a 90% SOC which is not too far off your graph but your graph doesn't show the right shape of power curve matching the battery discharge rate.