use codemirror to tokenize markdown

[erykwalder]

This is somewhat helpful, but tokenizing isn't really the hard part. The hard part is the structuring of tokens in trees based on the rules of spacing and so on.