Code review comment for lp:~vanvugt/mir/no-SurfaceBufferAccess

(my 2 cents on cleanup-strategy)
It is a good point that ms::Surface is a big gob of functions that needs some refinement. I guess I'd rather have input::Surface, frontend::Surface, and scene::SurfaceBufferAccess remain distinct then subsume them all into ms::Surface because at least those 3 distinctions give us some hints at what various parts of the system want in order to function.