The slews you see in AltAzSimple/AltAzMount drivers come from the way Roger (who wrote those drivers) implemented tracking: every second, perform a goto to the foreseen position. It was mainly aimed at visual observing. It explains why plate_solving alignment aborts as it detects these gotos.
Now the question is what usage for an altaz skywatcher driver (driving AZGti, AZEQ, virtusoso and all other altaz mounts I don't know of) ?
- visual: should be ok, but never heard a report here from users if any;
- planetary imaging: did not test myself, @Dirk tried to do so, but it seems he has serial port communication issues and I can't help him here. I'm not sure you need plate solving for planetary imaging, even 'tracking' is usually performed when you stack the images, but I am not used to planetary imaging;
- deep sky imaging: there you need everything, as you state in your post.
As I already said, I won't add a button in indi-eqmod for alt-az mode: first it's risky for users, and it will actually add a bunch of if/then/else in the code. It is not as simple as adding a layer between alt/az and ra/dec. Tracking would use variable speed adjustments, and you have to compute period values and make timer callbacks at a rate which depends on the stepper resolution of your mount. And you may have zero period sometimes (infiinite speed at zenith or pole, I don't remember), hence what to do there ? In Equatorial mode, tracking is just starting the RA motor... Concerning guiding I don't know if guiding programs may use Alt/Az axes, hopefully they do.
In other words, this represents a huge amount of work, refactoring indi-eqmod code first to split it into eq and altaz mode, refactoring alignment too, and then testing all that new stuff, which is at least half of the work. I still have a pair of pull requests waiting for users to confirm. AzGTi snap port is one of them, on my side I only use simulator mode, put A5 as mount code, see the snap button appear in the control panel and toggle it, but I still don't know if that actually works in real. What I mean is that I need testers, merging a 3 or 6 month branch may be time consuming (or not). And sometimes users simply vanish (that was the case for snap ports in indi-eqmod, I made the work twice).
Thus my point of view is that I don't plan to invest time to develop an altaz driver as there may be other solutions (which may need tweaking, but without testers...). And why not using an equatorial wedge if you want to plate solve? I don't know when does field rotation appear in altaz mode, but plate solving may also fail in altaz mode for that reason, even with a perfect smooth tracking. Or am I wrong ?