Code review comment for lp:~kdub/mir/fix-1299977-take3

I started on a proper solution for bug 1299977 back in May. That involves separating the concepts of window size and client size (as you'll find all existing windowing systems do). Once that's clear, we just switch the occlusion test to use the extents (including frame) of a window instead of its client region.

So the right (and elegant) solution is coming. I'm concerned about any change that appears to fix bug 1299977 without implementing this. It's suspicious but I haven't done a review yet...

If the only goal is to resolve bug 1299977 then I suspect this is probably the wrong answer.