Resume always starts from scratch if there are unicode filenames

[rsomeon]

f7e83a7 broke image resuming for me. n.decode('utf-8') raises an error about ascii codec and it's silently ignored, so nothing ends up being added to the list of downloaded files. I have a tentative fix: rsomeon@b7023a8. I have no idea about all this unicode stuff, but at least with that approach it shouldn't be worse than it was. Also, I'm not good with github, so two different things ended up in the same pull request: #249. :D sorry about that.