thanks Ellis, sounds worthwhile. a whole lot of images, but given the inacurracies of AF for many (most?) lenses and many cameras (Roger's lensrentals.com AF test is very interesting and consistent with what i have seen but not understood) this is probably a minimal sample to get results more useful than just doing it by eye.
it's still important for the user to decide which aperture should be optimized for lenses with significant focus shift like the Canon 50 1.2 and 1.4 - and then realize that the same focus uncertainty applies with AF in any image (probably only noticeable at larger apertures).
i have a theory that hand held, AI servo and multiple images should be shot with Canon cameras for highest probability of at least one critically sharp image (hoping there will be re-focus from image to image). haven't yet figured out an efficient test to verify this
of course, if you're using a tripod for still subjects, forget the whole thing and use liveview and CD AF, or even better magnify and focus by eye
Unfortunately, Canon and Nikon have a fair amount of catch-up to do on lens servo tolerances (and probably design) before PD AF matches the capabilities of the latest high resolution cameras