The 40 FPS on the R6 Mark II is limited to a burst of 2 seconds, while the 30 FPS on the R5 II and R3 is pretty much unlimited till the card is full, same goes for the 40 FPS on the R1. So the truly continuous / unlimited 30 FPS is far more useful to capture sports and action.
Where are you getting this data from? Are you confusing the burst raw mode witht he ES HS+ buffer?
According to my R6ii, I seem to be able to be able to churn through 40fps until the buffer fills up, then it's only slowed because it's writing to the card. On my camera and card combo, that's about 90 RAWs (2.5 seconds), or about 120 CRAWs for about 3.5 seconds. Which I belive is about the same as the R5ii. The original R5 had a longer 7 seconds HS+ duration in ES mode because of the 20fps limit. If I dial inthe same setting on my R6ii, ES HS (20fps) RAW, I seem to be able to shoot for around 12 seconds before the buffer fills, around 250 raw files. The data through put is roughly the same for all three cameras. it's jus tthe internal buffer size that limts the time duration before the camera needs to dump the files to the card reader. If you dial the R5ii or the R6ii down to 20fps you get more shooting time because the buffer is filling slower. What we are seeing here is the camera's internal buffer maxing out before the camera tips the files onto the card reader.
According to Duade Paton, the R5 and R5II have the same buffer size. The R5II has around 90 RAW frames of buffer in ES HS+ (30fps) mode. if you dial it to 20, you get more. CRAWs I belive the buffer is around 230 shots before the buffer maxxes. Whihc are the same figures I'm seeing on my R6II with the same settings dialled in.
The R3 gets 30 fps for 150 shots, which is 5 seconds by my math. I can't dial in 30 fps into my R6II. My Maths makes that R3's buffer about 1/3 bigger than the R6ii / R5II's buffer. Maybe Canon should invest in some bigger buffer chips than the OLED screen?
On the R6ii there is a pre capture and Burst mode that allows a 0.5 second (20 frames) of shooting combined into a single raw frame to be decoded in post production.