blog feeds: parse URLs from both description and content

[robx]

Currently, pzv links from an article description won't be seen, e.g. for errorfunction.tumblr.com (compare #12), which turns out to have a description but an empty content.

The most straightforward solution would be to fall back to description for the content field. I expect there to not be cases where description has puzzles, and content is non-empty but misses those.

[robx]

I've made this change, but haven't put it live on the database yet because that's a bit painful given the current structure.

[robx]

Fixed.