The crawl ledger: every page the desk crawler knows about, richest first.

Ordered by richness descending because that is the order the crawler keeps them in — the top of this list is what the Knowledge Document was compiled from, and human-pinned pages sort above the rest for the same reason. NULL richness (a row discovered but never successfully fetched) sorts last.

Always 200s for an accessible domain: a desk that has never crawled returns an empty ledger with has_provenance false, which is a legitimate state and not an error.