In RFC6381, there's no rule stating that the first part of codecs should
be video and the second part should be audio, while it seems the case
for data reported by YouTube.
In some dailymotion videos (like http://www.dailymotion.com/video/x3k0dtv from #8156) the segments URLs don't have the 'Content-Length' header and HttpFD sets the 'totat_bytes' field to None, so we also use '0' in that case (since we do different math operations with it).
Add an IE to support bigflix.com. It uses some sort of silverlight
plugin whose video url is being populated using base64 encoded
flashvars. So it is quite straightforward to extract.