Wrap map draw in camera.begin/finish so world-space transforms apply
Without the camera scope, maps.draw_map() rendered the map at screen coords starting at (0,0) regardless of camera.set_target, and screen_to_world's inverse produced bogus cell coords (mouse in screen-center mapped to top-left tile). Spine-prototype's render() shows the right pattern: world-space draws inside camera.begin/finish, screen-space overlays after.
This commit is contained in: