[Klaus]

Nothing you outlined explains why FF lenses are reacting worse to the procedure than lenses of a smaller format.
USM has the inverse effect on the numbers as an AA filter - no more, no less.
USM does, of course, boost the center because the center is always better - and it does so regardless of the image format.
It still doesn't explain why, on average, FF lenses are less receptive to USM at the borders than smaller format lenses e.g. the MFT tests have roughly the same pixel density as the EOS 5Ds R tests.

To throw you a breadcrumb - of course, if there was a MFT sensor with the same output as a EOS 5Ds R, the border results would tank massively compared to the center as well.
That just happens if you throw more and more sensor resolution at lenses that don't improve as fast as sensors do.
Megapixels simply follow the rule of about diminishing return of investment. So the different formats have different reasonable megapixel peaks.
So if you'd like to correct me, it's not about the lenses being crappy, it's about throwing too many megapixels at them.

Where this "reasonable" peak resides is subject to debate, of course. But for FF it is not 4x as high as for MFT, it is less than that.
This is all no news anyway. Medium format lenses are typically less sharp than full format lenses as well.
It's also the reason why smartphone images aren't quite as terrible as you might expect just from the sensor format and those tiny lenses.