It is the reason you are able to enjoy the accuracy you speak of... Without sync:
Slew to position A, mount slews with error, but thinks it's where it should be. Plate solve shows it is not where it thinks it is (need to sync here to introduce the delta between actual position and position A, but in this scenario we we don't), we issue the slew command again, mount does nothing because it thinks it's already where it should be.
To be 100% honest, we don't have a huge interest in refactoring the centering process to work with mounts that don't allow us to sync (not saying we won't do it either). A syncless slew can certainly be done if we take the burden of nudging the mount around, but nudge does not work for all scopes either. If this becomes the number one chief complaint (or at least is high in the list), we can always consider it...