Fully rewrite the HTML parser

[aecio]

Current implementation is messy, very hard to maintain, and make changes. New implementation should be compatible with current one and add new features:

• [x] Should normalize relative links
• [x] Should validate links and discard invalid ones
• [x] Should extract deep web .onion links
• [x] Should extract anchor text
• [x] Should extract text around links
• [ ] Should extract meta-tags (description, keywords, etc)
• [x] Should decode HTML entities to regular characters (turn & into &) from links
• [x] Should decode HTML entities to regular characters (turn & into &) from text
• [x] Should remove the fragment portion of the URL (anything after the character #)
• [x] Should do basic link normalization (lowercase domain, reorder query parameters, etc)
• [ ] NEW: Extract links to images and regular links separately
• [ ] NEW: Allow for easy extensions such as extraction of meta tags such as og:description, og:title, etc
• [ ] etc