No, it can't be X because 120Hz refresh rate is in the official Canon specs (or Sony A9II specs).
At best this implies that throughput is 120 fps, and at worst that the EVF refresh alone is 120 fps. (And in fact it's probably best and worst for the same camera depending on light levels.) But either way this advertised figure guarantees nothing about latency and nothing about the size (time) of the bubbles that are introduced by still capture.
And I don't think it's a misleading figure because the cameras can do it with minimal delay.
And that's a naked assertion which goes against observation. People are telling you they can see EVFs fall behind during burst shooting. We'll see how the R5/R6 hold up, but this has long been a problem with mirrorless.
'Latency' in this system, as above, is simply the processing time plus readout, and it physically can't be longer than the refresh rate...
It most certainly can. The path from exposure to EVF is a pipeline with discrete steps. No single step can be longer than 1/120th (assuming a true 120 fps in sufficient light), but the total path can theoretically be any length of time. That's the difference between throughput and latency.
'Delivering a frame to the display buffer' is not a thing, the whole processing is done right in the EVF buffer so it's ready right after processing is done.
It would be detrimental to performance, and difficult to code, if the processor had to interleave reads/writes with display reads. It's far easier to synchronize a single write of a completed frame against the display's 120 Hz read cycle. That EVF has its own discrete display buffer. There's no way DIGIC X is contending with the EVF for access to memory, it would just slaughter performance/efficiency.
Not sure what bubble you're referring to.
During the time it takes to readout a full resolution still frame and reset the sensor, no data is being fed to the EVF pipeline.
During continuous shooting, the EVF will obviously stutter but still able to display the latest captured image after each exposure.
Again, some cameras observably fall further behind during burst shooting. But even if the R5/R6 do not fall further behind, 1/20th or 1/12th + latency...which could be longer when processing a full resolution readout to the EVF...can be an eternity for some action sequences.
At the end of the day all that matters is the photographer's experience and performance. There's stutter on the highest end mirrorless bodies. Can you work around it and track any way? Sure. Is OVF blackout less disruptive? A lot of people think so.