There is an inherent overshoot or undershoot in the RA axis, simply because by the time the slew completes, the stars have moved - the RA tracking is effectively disabled while slewing. This will affect all mounts controlled by EQMOD, not just AstroEQ. EQMOD corrects this by performing a short slew at the end to get back to the target.
What's likely happening is that the acceleration table for your specific settings is some edge case that I haven't tested. Each mount settings will have a different acceleration table to try to achieve linear ramping. To allow for deceleration, the go-to is stopped short of the target by the total number of steps that it should take to decelerate, thus theoretically ending the ramp at the target.
However this process is an open-loop control (no feedback). If there is a mistake in the table, your slews can potentially keep going for a litter or a lot longer than intended as the mount takes many more steps to decelerate after the go-to completes than intended.
If this occurs, you will likely see lots of oscillations as the mount overshoots, then EQMOD tries to correct, but it overshoots again, EQMOD tries again but it overshoots, etc. It would either eventually settle on the target, or EQMOD will give up trying after I think three correction attempts.