Improve charset handling

[BjoernPetersen]

Loading m3u playlists from a file requires the file's content to be decoded using a Charset. Currently the user can specify a Charset for decoding, otherwise the parser defaults to UTF-8.

Instead of expecting the user to choose a Charset, the parser should default to the local system's default non-Unicode encoding for .m3u files and UTF-8 for .m3u8 files.

As always, we can't expect all .m3u files to follow the "convention" of using the local system's encoding, so we should detect signals of having used the wrong Charset and then fall back to UTF-8 anyway.

Users should still be able to completely override this "auto detection".