John, your two lazy susan tests cover the areas of interest. The "directly-mounted" test reflects on-ground behaviour -- what could be expected during take-off. The results also reflect what is observed in the real world -- substantial rudder input requirements for tracking during take-off.

The suspended test is representative of an in-flight situation, & again, the results fit with the real world -- very little effect in flight at low AOA, as would be expected.

I think that your empirical tests strongly support the spiral slipstream theory.

Is it possible to conduct a test at a positive AOA? You may have a powerfull fan available to add low-velocity free-stream airflow (near stall conditions). You would probably need some flow-straighteners to approximate the real thing.

This test might adequately demonstrate the add-in the effects of P-factor.