this extends zsh's _urls completion, allowing it to complete urls from html
pages.

the parsing is very simple, taking only href="(XXX)" into account. this
approach is entirely sufficient for simple file listings, but will have
problems with more complex pages.

Screenshot: http://mugenguild.com/~valodim/zsh_httplisting.png

Installation: Just put the _urls file somewhere in your fpath. You may need to
rm ~/.zcompdump and restart your zsh to refresh some completion cache.

TODO:
 - handle absolute urls (just ignore?)
 - do testing for different scenarios
 - do bookmarks and this local server thingie still work?
 - fix bugs?