We're checking we can't read an arbitrary file when we set each file:// URI to have a unique domain.