riccaruf Asiair Pro plate solve that seems to me works really good.
If the ASIAIR Plate Solve is working for you, you could try an experiment when you have a clear night.
Point your mount at any star field, near to the meridian. Do a plate solve and note down the RA and Declination. Now move the mount's RA using the ASIAIR arrow keys away from the Meridian (so you won't cross the Meridian). Just 10º to 20º should be sufficient. Now do another Plate Solve.
If everything is good, the second plate solve should show no declination movement since there is no declination motor to move. If there is some change in Declination, you can repeat a few times to see if the declination error is random, or if it is consistent. That should give you more idea of how your mount/OTA is behaving. If an RA movement causes a declination movement, then you may need to hold off using dithering in ASIAIR until ZWO fixes the bug, or switch to some other software that knows how to ignore the declination error.
Remember that even when you don't slew the mount manually, it is being slowly slewed automatically to track the movement of the stars at sidereal rate. So, there is always an RA movement. So, another thing you can do is to do two plate solves 30 minutes apart (the mount should have tracked for 7.5º) without touching anything in between. There should be no declination change. Any change is either coming from inaccurate polar alignment, or mount problems.
You really cannot completely avoid the problem with hobbyist equipment. So the real solution is for ZWO to ignore declination errors.
I did (imaging, graphics and color science) algorithms development myself before retiring back in 2005. Still write software for fun today.
Chen