When I changed from 5DII to 5DS I compared both that and the R side by side, and I agree with you 100%. IMO it's a mistake to compare the 5DIV to the 5DSR and then from the IV data try and interpolate what you think the 'sharpness' of the 5DS would look like, because the AA filter on the 5DS is much weaker than that on the 5DIV. To match the sharpness of the S to the SR I found (using PS USM) to be in the region of 80% of 1 pixel, which is minuscule sharpening. Also the fact that the 5DSR does have the same AA filter as the 5DS but 'cancels it out' by passing through another reversing filter would rather suggest that if it's possible to reverse the effect of the AA filter you aren't going to be able to recover more detail than it allowed to pass through ! Ergo a minute amount of USM does the same thing.
Also the 5DS is very sharp in it's own right, and shot noise (photon noise)can become quite evident. I presume that this is because in the raw file there really is no noise reduction if you switch it off in the raw converter. This doesn't appear to be the case in say the Nikon Z files, which are suspiciously lacking in shot noise, yet according to Bill Claff of photons to photos there isn't any noise reaction applied either......compare some of the test comparison shots that some websites have and the Canon is definitely a little more defined. Odd. I'd love someone to explain that one to me.
When it comes to a second body, IMO the best one is somewhat genre specific. For weddings / social events I'd always go for two identical bodies as you're justing both equally together.
Another interesting thing, at least here in the UK, is that although the Nikon D800e held its premium on the used market over the D800, this isn't the case with the 5DS / R, where here the 5DS can actually be more expensive used.