So I went back to the plastic guide I had made. Compared with the aerofoil:
You'd think that if the aerofoil were acting just as a simple guide due to its curvature, the plastic guide would work as well. But it doesn't*. I did further back-to-back testing and the plastic guide does basically nothing.
For testing it works really well having the wool tuft whiteboard marks on the inside of the glass: I put the aerofoil back on and the tufts returned to exactly their previously marked positions.
I then trialled just one 290mm long aerofoil and its
nearly long enough to get the good behavior when it is positioned liked this:
The rubber vortex generators semed to work best on the low window tufts, so I may be able to use a mix of the 290mm aerofoil and the rubber vortex generators to get the attached flow I want.
(* I am starting to think that the
Edgarwit's excellent performance is at least in part because of their true aerofoil shape. If you think of an aircraft wing's downwash, that airflow direction will help the Edgarwits guide flow back to the car body sides, and of course the same applies to these A-pillar air guides.)