It would seem you totally missed the meaning of regression testing. You don't seem to realize that you could measure the coefficient for force that goes as velocity squared (aero drag) and the coefficient for force independent of speed (rolling resistance and frictional losses) without keeping power, atmospheric conditions, or wind constant (as long as they're known), and then characterize the performance of each bike.

The precision of such indirect measurements would most likely invalidate any results.

... for all but the most extreme changes, power production is limited by physiologically not position

Well the change of position was pretty extreme considering how much steeper the equivalent seat tube angle was. And since I'm pretty new to cycling (my second season) it could affect my power much more then that of seasoned person with proper fit and technique.