feat: Wikidata batches and Commons pages for Hardy

design: Hardy volume PDFs and Wikimedia records

feat: OCR as per-page markdown, and the frc corpus ingest

implement the four-layer schema: scans tool, migration, JSON Schema

design: pages are always inline

design: schema for primary sources and scans