Code review comment for lp:~brandontschaefer/mir/update-confined-region

I agree, attempting to do a surface observer in the abstract shell has proven to need a bit more thought. Its strange having to store a unoredered mapping of the observers for a surface id. As well as passing the session/surface id/seat to the observer so it can check if the surface is confined before updating that seat.

My other thought is to possibly add a new callback to the observer update_confinement_region(geom::Rectangle const&) ... or Rectangles (as the input bounds).

More thinking to come here (would like it a bit cleaner then the current branch im messing with WIP until i figure a better way)