Cool we have the same train of thought lol. It all comes down to a bunch of "correction" factors to make that vague force reading from the strain gauge actually mean something.
Plus that computation of excess torque over the drag of the dyno....but it really can be made accurate if one goes to enough trouble and understands the equipment.
Probably where things go off the rails is that folks don't want to pay for some calibration service to come and set up to really put a know torque on the dyno. It would not be cheap or simple to do it brute force way..... a new 500 HP electric motor costs in the range of a few tens of thousands of dollars, and weighs 1-2 tons. And would take one heckuva power connection!