Fixed empty return in top_meta_image

[0x0ece]

Return empty unicode in ContentExtractor.top_meta_image() when no image is found. It was returning the article's url.

Example:
http://www.usnews.com/news/entertainment/articles/2015/01/15/boyhood-keaton-moore-are-oscar-favorites-in-vegas
(note that USNews often updates its articles, e.g. by adding an image -- at the time of writing it was without images)

[codelucas]

Nicely done, my only comment is that newspaper on master is a python 3 library, so return u'' should be return ''.

Don't worry about it though, I'll merge this on the command line and apply that tiny fix before the code goes in later today.

[0x0ece]

Ops, sorry about that and thank you!