Worked on mol-includes all day and I'm now at a version that is at least coherent, correct, and complete. It still needs to be rationalized, but that's a larger project (that may be in conversation with TEH's recent work on the teiHeader). Also added a schematron rule for local pointers; the only thing holding us back before was that it made Stow a pain to validate (another reason why splitting Stow is the right idea).