I definetly don't understand your discussion. In some cases you seem to agree but in the end you disagree. That is confusing.
I think we agree that for the same chip size, as you increase the number of pixels (or decrease the size of the individual collection sites), for each site as the density increases, the individual site collects less light. So, Canon in increasing pixel count, actually decreased the ability of each site to collect light and decreased the signal to noise (made the noise performance worse) of each site. If for example, Canon would conversely chose to go to 24MP, then each indiividual site would be larger and collect even more light than the smaller individual site on a 32MP chip; and the signal to noise would increase (noise performance would improve at 24MP over 32MP). So it is incorrect to say that simply going to a larger format, is the only way to improve noise figure, if that is what you are implying. I do also agree that as you increase the size of a photo site by going to a larger chip size or format you should also improve signal to noise but the discussion within a specific chip size and the fundamental underlying physics is the same. Increase site size, better noise performance, decrease cell size, worse noise performance. There are design choices that can be made to make the signal to noise better on an APC-S chip.
If the rumour is true and Canon has decided to go to newer sensor technology as in BSI, then that technology actually improves each sites ability to collect light. BSI technology moves the collection site higher and closer to the lens actually improving light collection and increasing (making better) the signal to noise.
If Canon stays at the original 32MP density and improved the sensor technology by evolving to a BSI sensor, then the noise figure could have actually improved on the new R7 APC-S chip.
Now the problem I have with all the "rumours" or speculation is that it was suggested strongly that this MII was targeted for action shooters. If that is the case (and I now question if that rumour was true), then they should not have increased pixel density (it wasn't necessary) at the cost of the same or worse signal to noise in the camera. The R7 is well documented to be noisy at ISO 6400+. The R7's noise performance could have been made better simply by improving the sensor technology to a BSI chip. However, and this is what is confusing about the rumour or marketing, it seems Canon is rumored to also want to support 8K video, requiring a higher pixel count to do so. If you are a photographer, what you wanted to see in a Canon R7 MII was a reasonable resolution (like 32MP). better high ISO performance (like at 6400 for those many moments where there is low/poor light and you need shutter speeds of 1/1000 for action photography - helped by a BSI chip), a reduction in rolling shutter (also helped by the addition of a BSI chip with faster readout), some reasonable improvement to the number of frames per second, and perhaps better pre-capture. You can technically accomplish those goals on APC-S chip and I contend it is what most photographers wanted to see. If however, you want a device for 8K video, than perhaps either there should have been two versions as in the R50. I'm not interested in the 8K part, if it compromises the actual image quality that could otherwise have been had.