While I have not dyno'ed many Spec Miatas, I have either for personal cars or cars through my shop dyno tuned larger V6 and V8 cars( both full race and street rod types) and feel that 1% error would be acceptable/expected and the 1 to 2 HP differences between pulls to be "noise". Maybe Drago or Kessler can say that in their experience,if they did see repeatable 1 to 2hp differences that translated into repeatable results on the track that may help others to get their head around this, but at the regional level,would the majority of SM drivers feel the difference? I feel, and a few well known race engine builders(albeit circle track) feel that a chassis dyno just isn't as "scientific" as an engine dyno. There are just too many variables(noise) that can affect repeatability/accuracy to worry about 1 to 2HP. I may be out of line here, but are we maybe getting crazy over numbers that may or may not mean anything?
1-2HP differences can be observed on consecutive pulls with no changes to anything when dynoing a Spec Miata on a chassis dyno. I have hundreds of files over the years and have seen this pretty much every time. Do a run, record the data, do another run, record the data. Do NOT expect identical results, they will typically vary between 1-2HP.
I personally doubt that the 1.5mm allowance will provide anything more than 1-2Hp increase if that much.
With regard to the repeatability of Chassis Dyno's they are nowhere near as accurate as Engine Dynos where there is much tighter control on the ambient conditions. However there are very few engine dyno's currently setup for SM. I only know of one. I doubt that the testing will be done on an Engine Dyno, but I have been wrong before.
I believe that NASA/SCCA/Mazda is currently more interested in determining what the 1.5mm STR allowance equates to in terms of performance gain.