Normally dynos are done in the 1:1 gear ratio. For instance, if you dyno a 300ZX twin turbo in 4th gear, the ratio is 1:1 so the torque at the shaft is the same as the torque at the wheels. The Tesla obviously can't do that because it's stuck at it's final drive ratio. Making it even more complicated, the rear is 9.73 and the front is 9.34 so if you're going to calculate torque at the shafts, the best you can do is split the difference:
Here's a vbox run with torque calculated at the shafts:
Here it is again but at the wheels. In this case, we don't have to be confused by different final drive ratios between the front and rear motors:
Notice how torque is a function of g's.
As far as I know, the two motors are geared differently, so unless you measure and report them independently, "at the wheels" is the only way that is correct - the Mustang AWD dyno from the first link mechanically connects the rollers to keep the speed constant, and thus only outputs one number; not sure how the others were done.
Walter
The Mustang dyno indeed only has one PAU and physically links both front and rear drums together with a toothed belt. The Dynojet has independent rollers with two PAU's but you can't normally test AWD cars in that configuration because the transfer case could allow the front and rear axles to spin at different speeds. So why does the Dynojet have this setup? Because you can't test 4WD cars on a dyno that has a linked drums without slippage for the same reason you can't drive a 4wd vehicle on solid slip free ground. You can only drive a 4wd vehicle in snow, mud, sand, gravel, etc. So if you want to dyno a 4wd car you need to use a Dynojet. If you want to dyno an AWD car on a Dynojet, then you need to use the Linx system which is belt that ties the front and rear drums together. In this configuration, each PAU no longer tests each drum independently.
If the Mustang dyno operator is trying to get torque at the shafts, they really better know what they're doing when calculating final drive ratios when they are aren't the same. In this case, because they are very close, it doesn't make that much difference on torque even if they use only the front or back. Or they can just use the wheels which I agree is the right thing to do in this case.
Regardless of how you calculate torque, the hp calculation is unchanged.