While I've loved using my RF 15-35 f2.8L, I'll admit I'm new to using any lens wider than 24mm and so I'm just happy to explore the new horizons the wider view gives me. I haven't tried to take images at MFD yet and I haven't been trying to critically analyze the edge & corner performance. I suppose I could try some tests and get back to you on it, but I don't have an R body to compare it with.
So your RF 50 1.2 & RF 70-200 2.8 look the same on the R and R5? That would indicate to me that your R is working as expected and there's no big difference in sensor technology regarding how they work with RF lenses. If you're handholding your RF 15-35 2.8L then I would expect the R lack of IBIS could be a major factor causing trouble, since lens IS can't correct for some of the modes of image shake, such as rotation while the IBIS can cover those in addition to improving all the other ones. I would suggest that you don't make any tests handheld and only test with a tripod and non-moving subject.
I would assume that the angle of incidence of the light hitting the sensor edges is getting very extreme as you get wider and wider views, such a 15mm or possibly at MFD (I'm just guessing there) and thus that might indeed show there is an effect of the microlens & sensor construction between the R and R5 for severe off-center angle of incidence. To better test that theory, I'd suggest you find someone else with an R body to test your 15-35 lens on to see if it has the same effect (only towards extreme wide angles) with all your RF lenses.
(ps - it's soooo nice to have calm discussions again)