Hmmm - never thought of that one.
The propwash vortex certainly reduces AOA downwind of the descending blade, and increases AOA behind the ascending blade.
Let's assume right-hand rotation of both engines (i.e. a typical twin). The spanwise CP is thus moved inboard a bit on the right wing, and outboard a bit on the left wing. (I think that's right...) With both turning, there's a bit of roll to the right, which may or may not equal the left-hand roll reaction due to prop-shaft torque.
Now if the left prop stops turning, the CP on that wing moves inboard a bit, reducing the right-hand roll moment. But P-factor tells us the left engine is critical, because the thrust vector is outboard of the RH prop centerline.
So I guess the P-factor issue is mitigated partially by the accompanying wing spanwise CP shift. (Honest, I've only had half-a-beer, Yank brew at that...)
