Use XML parser library

Currently parsing of the rendered RDoc or Markdown is via simple text processing. This should eventually be changed out for a real XML parser, probably Nokogiri, which will make the code more robust as well as more flexible.