> My question is this: at what point, if at all, will self-driving get good enough to make automotive lidar redundant?

By 2018, if you listen to certain circa-2015 full self-driving technologists.