Can someone explain to me why Tesla didn't bootstrap itself with cameras/LIDAR, then remove the LIDAR? I mean teaching an AI depth perception and the ability to model objects by vision, without including training data from LIDAR just seems kind of silly.