Valid HTML may contain unescaped > characters in text nodes

[DanielHeath]

Currently, Upmark errors out attempting to parse HTML which contains a > in any text node, which every browser supports.

Nokogiri::HTML5 will even generate such HTML without escaping the >.

For instance:

Upmark.convert <<-HTML
  <div>
    Foo => bar
  </div>
HTML

[tiagoamaro]

Hello @DanielHeath, could you add some tests scenarios for this new use case? Thank you!