R N Clark finds
here that his definition of picture quality (AIQ) is best with a pixel size of 5 to 6 um (taking the simple view that size = pitch, which he does address elsewhere). The implication is that , by his definition, the ideal MP for any camera is purely a function of the sensor size, actually it's area.
For example, a 24mm x 14mm APS-C best has about 11MP and a 36x24 full frame about 28MP.
A
follow-up to that link shows example images.
Clark makes the most compelling arguments, IMHO.
Ted