Code review comment for lp:~amanica/bzr/mv_after

This looks ok to me but I wouldn't mind a second opinion. My main concern is that we've been moving away from methods that mutate inventories (like inv.add()) in favour of generating new ones via create_by_apply_delta(). I *think* the code is all ok in this context but I'd like John and/or Robert (say) to confirm that.