Seemingly lists.cam.ac.uk no longer wants &#...; entity refs, and instead just wants real UTF-8

Switch from lxml to html.parser for CSRF token extraction, as bs4's lxml support on Ubuntu 16.04 throws a DeprecationWarning

Mailman 2.1.15 added a CSRF token on admin forms so include that in the request if needed. (Added bs4 dependency.)