Add document type support for everything we supported so far

This is largely existing code that previously was either in the generic
extractors or somewhere in ExtractorEngine, refactored to have everything
related to a specific document type in a single place.