Would a simple solution be to use a second ASCOM Safety Monitor to determine when it is OK to image?
Keep the existing Safety Monitor implementation to force the end of sequence, mount parking and slaved observatory closing when conditions dictate.
Add an additional ASCOM Safety Monitor instance to control imaging. When this monitor transitions to Unsafe, stop the current image, stop autoguiding and stop the mount tracking. When it transitions back to Safe, select a target based on time etc, slew & centre on the target, start guiding, run autofocus, and start imaging. This looks very like at least part of the start of sequence logic.
These actions could be repeated many times through the night without the need to terminate the sequence.
I have an AAG CloudWatcher/Solo. I would use Chris Roland’s Boltwood Safety Monitor as the existing SGP Safety Monitor, allowing the dome to remain open when cloudy, windy, wet etc, but close it whenever overcast, very windy, rain was detected.
I would use the AAG CloudWatcher Safety Monitor driver to allow imaging when dark and clear.
Equipment would be safe based on the current SGP use of the ASCOM Safety Monitor, and the fact that tracking would be stopped whenever imaging was paused. No complexity in meridian flips while paused, etc.