Before the arrival of SWG mounts, this was never really a question. PHD2 only generates max-move guide pulses when the tracking error is abnormally large, when large dithers require it, or when Dec backlash compensation is being used. Reducing the max-move values has no effect unless one of these conditions arises. AFAICT this has become a question with SWG mounts because the RA tracking accuracy of some of these mounts can be quite poor which is also the reason for using relatively short guide exposures. So a long guide pulse, particularly in Dec, can allow the RA tracking error to become large because the RA tracking is being "left unattended" for exposure_time + max_move_time seconds. If you reduce the max-move parameters, you will induce under-correction in the guiding for the situations I mentioned - I can't picture why it would cause oscillation. Remember, all of this is serialized, you can't start a guide camera exposure until all the guiding commands have completed. You have to decide if under-correction is a price you're willing to pay in terms of recovery times for things like dithering or large excursions caused by something in the mount payload.
Regards,
Bruce