Cache entry content in db

[facundoolano]

• Adds a new content_full field and rename the previous body to content_short
• Stores the cleaned up article html in the db after the first fetch and reuse it on subsequent requests
• Adds a periodic task to prefetch the content from the first few entries in each user's feed
• Prefetch when pinning an entry.
• Moved the scraping related code to its own module.
• Updated the nodejs script to accept article HTML from stdin as an alternative to the url argument (I ended up not using this feature).

I made some experiments on running readability during entry sync (for the ones I already had the HTML available) and that proved to slow the task too much.