This wasn't an issue with film so surely it would be possible for the sensor to 'autoalign' ?
an issue with film, just not a very big one. Most 35mm film provided the resolution equivalent of about a 6-8mp FF sensor, give or take (& depending on the film of course), so it was pretty tolerant to slight discrepancies in film flatness. Contax's RTS III (& some Contax MF systems) had a vacuum system that would suck the film against the backplate so that it was absolutely flat, and folks were able to get a bit more resolution out of higher-end film as a result.
A modern high-MP CMOS sensor needs to be *absolutely* aligned and positioned in the plane of focus, so closely that it pushes the limits of the production techniques used to make our mass-produced camera bodies. This is part of the reason why we have AF calibration in our cameras now, so that we can compensate for manufacturing tolerance variance in the distance between the lens mount & the sensor plane.