20260903T1957Z data model docs structure

What. The graph data model doc (proposed; parts 1 and 2 settled with Nacho 2026-09-03) migrated to the D28 docs structure on feat/graph-data-model. Concepts and invariants now cite docs/domain.md; the relationships proposal is cited for history only and the attribute schemas via interfaces.md's registry section; rule 6 cites design/conventions.md. The invariant-enforcement map gains invariant 10. The doc carries the concept-to-column mapping (superseded_by to superseded_by_id, partial kind to entity_type). Part 6, ingestion tables, added as a stub so the conventions bind ingestion's tables from day one. The three open questions are GitHub issues #8, #10, and #12 (labeled open-question, per the rule Nacho set the same day) and the doc's section only points at them.

Why. D28 made domain.md the one place for definitions and moved relationships.md and the skeleton proposal to proposals/, so the doc's links broke under the strict build and its invariant map was one row short. The data model rules in the writing-docs skill require the umbrella doc to bind every module's tables, ingestion included.

Verify. make docs-build exits 0 on the branch; docs/design/data-model.md has a row for invariant 10 and a part 6 row; issues #8, #10, #12 carry the three questions; HANDOFF.md is gone from the branch root.