ID3 parsing fails for UTF-8

[michaelsanford]

Песни о Любви и Смерти again…

Maybe push it upstream, too, if we fix it.

[NebuPookins]

I think the first step here is to try and generate some reproducible test cases.

We should:

• Read the specs http://id3.org/
• Find or generate some free creative commons or other GPL-compatible mp3 files (e.g. just have an mp3 file of 1 second of silence).
• Edit the tag to contain unicode characters, e.g. http://www.ltg.ed.ac.uk/~richard/unicode-sample.html
• Write a unit test that makes sure whatever library we're using can correctly decode the file.

(We might need examples for ID3v1 and ID3v2).

Not sure, but I suspect there's no standard for what encoding to use for the unicode character (e.g. it might not always be UTF-8). If so, we might look into some sort of heuristic for "guessing" the encoding.