The RA axis actually visually move. Not by a lot, but much more than can be explained by sidereal.
I don't rely on the camera for anything yet since I can't rationalize the mount movement yet.
I've just looked up the RA/DEC coordinates as the mount report it when it's at those 2 Alt/Az points:
Alt 0, Az 90 translates to RA: 20:25:25 Dec: 0
Alt 0, Az 270 translates to RA: 8:25:25 (+ sidereal) Dec: 0
So delta the meridian flip, that's the same RA plane - at least I'm not completely going crazy. But it doesn't seem to travel on that plane from point A to point B when it's slewing...
This is not technically an issue for me, since I want to move the mount by hand anyway. The issue is that I don't understand why it's doing this during an automated slew, so I don't know if my tests are valid.