I've - oddly - owned three of those sigma 500 f/4s. The second one was purchased specifically for the R5 thinking that it would be a place-holder until the RF big whites came around. (The third was a mistake, as I wound up thinking I had lost the bid and got the 2nd one, but then found that I'd won both auctions). I've sold them all since.
The focus shift issue is significant if you don't shoot the lens wide open. I almost never shoot the lens at an aperture smaller than wide open, so it didn't bug me at all. The distance at which I shoot things tends to make the plane of focus deep enough to not worry about it. And stopping down doesn't improve IQ much, unlike the old days.
Anyway, the focus speed was just poorer enough versus the Canon 500 (which I'd owned previously) that it detracted. Took a great deal away from the R5's tracking capabilities. My sense was that this had more to do with the speed of the tracking motor rather than any algorithmic nuance. I didn't assume a firmware would be able to address it. Would be glad to be wrong.
After the Mark III version of the 600 f/4 came out, the Mark IIs came down in price on the used market, so I went with one of those. The equation was used Sigma 500 at $3800 versus a used canon 600 II at $5500. With the extra focal length and the AF competence (and IQ as good as the mark III), it was the right choice for me. Despite it's larger weight than the Mark III it might continue to be the best choice for me if the new RF 600 turns out to be a glued adapter job.