Add normalization of file names after creating .torrent file

[dramiik]

When my files contain non-ASCII characters (like diacritics, accents, symbols etc.), the .torrent file created by mktorrent doesn't pass the verification in my Torrent client. This is because of the difference in normalization standard on MacOS (where I create .torrent file) and Linux (where I store files and run Torrent client). For more details see pobrn/mktorrent#14.

I've decided to post-process the .torrent file and add a function to normalize included files. Even though I tested this on my setup with multiple torrents, it would be great if someone else on MacOS could reproduce and confirm having the same issue.

Side note: I also tried to replace mktorrent with torrentool which is written in Python (https://pypi.org/project/torrentool/). Didn't solve the issue with normalization.

[walkrflocka]

I like the direction you're going here, but it gives me pause that this isn't a narrower patch. I feel like we could handle this at creation time - admittedly I am not an expert on Unicode encoding, but there's got to be a better way than mutating an already created torrent file. Can you look into either setting up a narrower fix that checks for the presence of the NFD encoding rather than blanket assuming the torrent needs normalization?