Cleanup markdown parsing

[matthew-carroll]

Currently, the markdown-to-document parser creates document nodes in visitElementBefore, however, it would be more appropriate to log the block type in visitElementBefore and then construct the actual document node within visitText and/or visitElementAfter.