Process at markdown parsing.

[bobvanderclay]

Several markdown processors already have the option to generate the ToC, at least generate header id's.

It's clear the header detection and extraction would be better done at the markdown level.

Even if I used a markdown processor with ToC generation, I still want this plugin to extract header information to use in templates that might not be "near" the document. The added flexibility to template out the formatting of the ToC is useful, too.