Extract a file into the ontology (doc + space + org levels)

Extracts entities and relations from a file once and populates the document, space, and organization graph levels together. JSON body: fileId (required, non-empty). The file must belong to :spaceId or the request returns 404. Returns { fileId, results }, one IngestNamespaceResult per graph level with status (ok | error) and the entities/mentions/edges counts written. Idempotent — re-ingesting creates or enhances the existing graphs. Requires space:contribute on :spaceId; lacking write permission is a 403 (a missing space:read prerequisite would be a 404).