It would be really cool but it will require the video to have depth information to work. Probably will need to be generated with AI if it's to work on all videos and then streamed along with the video as real time depth estimation for the entire video is not feasible in real time at the moment on a quest.
So, not a small request but I agree, definitely a worthwhile venture. Especially as that depth information can also be used to create proper auto focus.