See if we can use ENCA for automatic charset recognition on subtitle files.
Might simplify usage of sub files a lot? Might require complete file to be parsed by ENCA before good result is returned