Use Selenium for browser-based scripts

[Chris Warburton]

Some of our scripts interact with Web pages. There are three levels to
this:

 - Nice sites: these don't rely on crap like Javascript, so we can grab
   their HTML using wget or curl, and do what we like (e.g. XPath, etc.)
 - Annoying sites: the things we want are only available by executing
   Javascript. This might not be too bad, e.g. if it's an "on page load"
   event, we can run it through e.g. phantomjs, wait a few seconds, and
   dump out the resulting page source.
 - Crap sites: these not only require Javascript, but they only seem to
   work in bloatware browsers like Firefox.

For the latter, we're currently launching Firefox on an XVFB display,
using xdotool to perform actions (like opening the Web inspector, typing
into the console and copying the source) and xclip to dump out results
from the clipboard.

This is horrible.

It's time to bite the bullet and try out Selenium, probably via the
Python library. If it's nice enough, we might be able to use it in place
of phantomjs too.