Hypothetically, of course, I could imagine that in mosaic mode there is additional time spent after each image acquisition to perform a quality check, alignment, etc. And then it proceeds to get the next image. That might add, say, 20s to a 10s exposures.
If, instead, there was an option for all the frames to be acquired first and then have that quality check, alignment and mosaic construction done after, the user would be able to take advantage of shorter windows of good seeing. True, there wouldn't be a live stack/mosaic taking place. But the user gets an image that they otherwise would have missed.
Just a thought ...