layer-shell: check if the surface is mapped in layer_surface_destroy()

If the layer surface has been closed by the compositor, using layer_surface_close(), then the unmap event is emitted immediately. However, when the layer surface is later destroyed by the client, the compositor used to get a second unmap, which is fixed with this commit.

As discussed in IRC, there could be another way to fix this - by not emitting the unmap on in layer_surface_close(). However, there is still the possibility the client first unmaps by committing a NULL buffer and then destroying the surface, so the change in this PR is needed anyway.