We currently have 16 endpoints. The IDs are typed by a human at creation time. This check will detect with we ever use an endpoint ID twice. Since the large irrelevant directories are ignored, this should be quick enough.
