This puts images referenced by HTML nodes below that node now, contrary
to the MIME structure, so HTML extractor scripts have access to images
(and their contained barcode content) that are referenced by cid: URL.