docs: open-issue priority + dependency DAG snapshot

[robotrocketscience]

Summary

Single-doc index for the open-issue tree. Sorts everything in gh issue list --state open into five waves with explicit deps, soft-deps, and cross-cutting hazards. Goal: prevent the rework patterns that already bit us (#293's tests dying on #283's UNIQUE constraint, #256 stuck on author rebase after multiple PRs landed on store.py, etc.).

Five waves:

• Wave 0 — close-out. Seven stale items that should walk to zero before new work starts (feedback_history has 100% orphan belief_ids; audit trail is structurally broken #223, [v2.0] T1 corroboration recorder trigger gap — content_hash UNIQUE constraint missing #254, Context rebuild surfaces duplicate + off-topic beliefs; ranking has no relevance floor or output-dedup #281 re-scope, Context rebuild redesign — scoping (parent) #286 → tracker, Rebuild redesign: eval harness — per-rebuild log + fixed-corpus precision #287 dup, hook hardening: UserPromptSubmit memory hook is an injection surface #280, Rebuild redesign: eval harness — per-rebuild log + fixed-corpus precision #288 implementation tail).
• Wave 1 — rebuild redesign (Rebuild redesign: eval harness — per-rebuild log + fixed-corpus precision #288, Rebuild redesign: silent-vs-always contract + relevance floor #289, Rebuild redesign: belief typing + aging policy #290, Rebuild redesign: query understanding (entity extraction, intent, term-weighting) #291). All four spec memos in flight; ratify all four before opening any implementation PR. Implementations must sequence (context_rebuilder.py is a merge-conflict spot).
• Wave 2 — phantom-prereqs T1→T2→T3 ([phantom-prereqs T2] Implicit retrieval-driven feedback — sweeper + grace window #191/feat: deferred-feedback sweeper — implicit retrieval-driven posterior signal (#191) #256 stuck on rebase; [phantom-prereqs T3] session_id propagation through ingest entry points #192 blocks on T2).
• Wave 3 — v2.0 substrate decision tree. v2.0 substrate decision: multi-axis vs single-axis uncertainty #196 gates everything posterior-related. Bench-gated items (Deduplication module (dedup) — v2.0 evaluation #197/Multi-LLM consensus module (multimodel) — v2.0 evaluation #198/Enforcement module — directive detection + compliance audit + selective injection #199/Semantic contradiction detector (relationship_detector) — v2.0 evaluation #201/[v2.0] Phantom promotion-trigger rule — three rejected naive triggers, need a benchmarked rule #229) wait on Rebuild redesign: eval harness — per-rebuild log + fixed-corpus precision #288's harness producing precision/recall numbers.
• Wave 4 — v2.x materialization ([v2.x] Implement replay_full_equality — flip-readiness probe #262 → [v2.x] Derivation worker — beliefs become materialized state #264 → [v2.x] View-flip — beliefs/edges become materialized; ingest_log is canonical #265). Sequential, must not parallelise.
• Wave 5 — long-tail retrieval / research ([retrieval] Pipeline composition tracker — unified retrieve() with feature-flag gate #154, [research] uri_baki post-rank adjuster retest with relevance-aware locked set #153). Hold; [retrieval] Pipeline composition tracker — unified retrieve() with feature-flag gate #154 refactor would rework Rebuild redesign: silent-vs-always contract + relevance floor #289–Rebuild redesign: query understanding (entity extraction, intent, term-weighting) #291.

Plus five cross-cutting hazards on the wall: context_rebuilder.py and store.py as merge-conflict spots; the #283 UNIQUE constraint test pattern; calibration-data bench-gate; substrate ratification before #290 implementation.

Decision asks

Bottom of the doc — operator stamping queue (5 items) that unblocks Wave 1 implementation in sequence.

Test plan

• All open issues from gh issue list accounted for
• All in-flight PRs cross-referenced in the spec / impl index tables
• Discretion grep clean
• Operator review: confirm wave assignments + hazards list