Custom charset support for subtitles

[tp7]

Demux/open/write subtitles with user-specified charset.

Something like this should work for FFmpeg (cp1252 example):

args.extend(['-scodec', 'copy', '-sub_charenc', 'cp1252'])

And then use cp1252 while opening the file instead of utf-8-sig.

This might not be actually needed as subtitles in weird charsets seem to be quite rare.

[tophf]

So here's some feedback.

1. "Not needed" only for modern anime subtitles.
2. cp1252 is/was a default charset for Western Europe so it's not "weird".

[tp7]

The ffmpeg demuxing issue is fixed in 1515e424bc1e6af7f2914a2c851a55eedf6bed14.

Now we need to figure out how to correctly open subtitles in different codepages in Sushi without any user input.

[tophf]

While the abovementioned commit is a sensible addition to Sushi it'll hardly be possible to correctly autodetect the encoding always (see this), although some (semi-?)solutions exist (chardet, UnicodeDammit).