Adding random filenames

[alexloney]

I had the same issue as the user in #1 where the URL I was trying to download from was too large, causing the automatically generated path for the temporary filenames to be too long and resulting in the error message:

OSError: [Errno 36] File name too long

To resolve this, I've added a CLI option of --randomfilenames which will cause all temporary filenames to be converted into UUIDs instead of their full names, thus all filenames will be shortened to an appropriately small length. Because of the way the m3u8 file was used with ffmpeg, the filenames need to be preserved and inserted into the updated m3u8 file, so this is achieved by creating a lookup table in the M3u8Downloader class for the filenames that maps the original full filename to the shortened UUID name, this way the m3u8 file may be reconstructed with a consistent random filename for the final combination of fragments into the full file.

[josephcappadona]

Looks good -- thanks for submitting the PR! :)