improvements to fxurl

[mildsunrise]

introduces a markdown parser so that the bot can understand the message formatting in the same way clients display it, and use it in fxurl.py.

consequences:

• only process links that discord recognizes as such (urls appearing inside code spans or code blocks are no longer detected)
• only process previewing links (if user disables preview with <url>, skip it)
• skip links inside spoilers

also took the opportunity to restrict the tests we make on the detected links:

• only consider http links (just in case)
• hostname must exactly match against a list
• for twitter, the link must be to a tweet or tweet photo (things like homepage, profiles, intents, etc. are now skipped)

I also added some normalizations for correctness:

• normalize escapes by unquoting/quoting
• case insensitive comparison of hostname

edit: rebased on top of the fxurl branch, which also brings the following changes:

• remove the explanatory note on fxurl messages
• WIP word filter

[vicky5124]

TODO:

• Masked links ()[]
• x.com support (fixupx.com)
• Translate twitter links to Catalan by adding /ca at the end of the path.
• Easily toggle between fxtwitter and vxtwitter in case of outage.
• Replace fxpixiv.net with phixiv.net