- path_to_uri() encodes unicode input as UTF-8 and leaves bytestring input
unchanged before it is converted to file:// URIs.
- uri_to_path() will now always return bytestrings, since we don't know if
there is any non-UTF-8 encoded chars in the file path, and converting it to
unicode would make such paths no longer match the dir or file it was
referring to.
- split_path() will now assume it gets a bytestring in.
b5d9dc10a7