Based on links that I've seen, /e/<videoid> also occurs in the wild, and making this substitution yields effective results.
Now it downloads the video over http in one file, it doesn't downloads in fragments Added a test and use the methods in InfoExtractor for downloading webpages