refactor(simulation): PR #137 review — audit override + regression tests + docs

Addresses Tyre, Hoshe, and lead review comments on PR #137:

- **Audit doc amendment** (Tyre E1 / Hoshe H1 / Lead): add "Lead override
  (2026-04-21)" section at top of docs/architecture/sprint-37-878-audit.md.
  Rewrites the conclusion to "DECISION: STRIP" with the cascade-based
  rationale. Preserves the original audit body as the pre-override record.

- **Regression tests** (Lead 2a-2b / Hoshe H2 / H3): add POSITIVE
  assertions of the new uniform behavior so silent reintroduction fails.
  - `phase2_container_verb_labels_uniform_regardless_of_player_state` —
    two trials (empty KG, POI-bearing KG) assert container verb labels
    equal Phase-1 defaults.
  - `monologue_pool_selection_uniform_no_archetype_key` — two observers
    with divergent MonologueState both draw from OBSERVE_NPC_LINES.

- **Decision record amendments** (Lead 3 / Tyre S2): D-032, D-035, and
  D-057 amended with Phase 6 deferral wording. "Retired pending Phase 6,
  not deferred with scaffolding." Reintroduction gate: a confirmed
  Phase 6 character-model design.

- **types.rs doc fixes** (Tyre S1 / Hoshe H5): StartupMessage protocol-
  flow comment updated to reflect no-version handshake (D-192).
  ObserverSnapshot version-history block grows a "Sprint 37 wire-format
  shifts" section documenting D-192 + #878 schema drops.

- **observer/tests.rs:944 comment** (Hoshe H6): rewritten to cite
  cascade rationale instead of the stale D-032-SUPERSEDED premise.

- **tests/run-atlas-determinism exit** (Hoshe H7): exit 0 when EXIT_CODE=2
  (venv/DB missing = skip, not fail). Preserves skip semantics for
  tests/run-all on machines without the Python venv.

Follow-up tickets filed:
- #895 (server, low): expand check-systems-db-stamp GENERATOR_SOURCES
  to cover gemma_naming.py + naming_core.py (Tyre S3).
- #896 (planning, low): add CLAUDE.md carveout for server wiki writes
  closing coverage gates (Tyre S4 / Hoshe H8).

H4 investigation: v01_integration_playthrough.rs was not the only E2E
handshake→tick→snapshot test; coverage preserved by bridge_ipc.rs,
bridge_tcp.rs, and game_loop.rs (the latter is pre-existing-broken
per #885). No replacement test needed.

1142/1142 lib tests pass. cargo clippy -- -D warnings clean.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

decisions/perception.md

@@ -225,9 +225,9 @@ How the player observes and interacts with the world: camera, fog, line-of-sight
   - `GameState.insert_active` is the source of truth (defaults true in v0.1; wired from snapshot field `insert_active`).
   - Rationale: diegetically consistent — the body reacts to proximity; the insert reacts to commands.
 
-### D-057: Entity interaction — vertical list, insert-styled [PARTIALLY SUPERSEDED]
+### D-057: Entity interaction — vertical list, insert-styled [PARTIALLY SUPERSEDED — archetype portion deferred to Phase 6]
 - **Date:** 2026-02-13
-- **Supersession note (2026-04-21, Sprint 37, #878):** The character-archetype verb variation portion of this decision is superseded by the development cascade (CLAUDE.md) — archetype-driven verb relabeling is Phase 6 detail-coloring and has been stripped from the server. Vertical-list structure, Phase 1/Phase 2 split, POI priority flips, and contradiction markers remain live. Relabeling (Open→"Move"/"Stash" vs "Scan"/"Flag") is deleted; verbs now render with their base labels regardless of character.
+- **Supersession note (2026-04-21 / amended 2026-04-22, Sprint 37, #878, PR #137):** The character-archetype verb variation portion of this decision is retired pending a Phase 6 character-model design — not deferred in place with a stub. Per the development cascade (CLAUDE.md), archetype-driven verb relabeling is Phase 6 detail-coloring and has been stripped from the server. Vertical-list structure, Phase 1/Phase 2 split, POI priority flips, and contradiction markers remain live. Relabeling (Open→"Move"/"Stash" vs "Scan"/"Flag") is deleted; container verb labels are now identical across all player states, and uniform labeling is the intended pre-Phase-6 end state, not a regression. Reintroduction gate: a confirmed Phase 6 character-model design is a prerequisite. See `docs/architecture/sprint-37-878-audit.md` (lead override section) for the cascade rationale.
 - **Decision:** Entity interactions use a compact vertical list (not radial). 2-4 options max, anchored to entity position. Insert-styled with Araminta's geometric aesthetic. New options unlocked by knowledge changes are highlighted with a gradient glow background. Radial menu reserved for world menu only ([D-058](#d-058-world-menu--radial-4-spokes)). Max 3 visible response options in dialogue context.
 - **Server architecture:** Two-phase verb computation. Phase 1 (simulation, no KG): compute maximum possible verb set from ObjectType component (Readable, Container, Terminal, Door, Pickup, Furniture — each with specific verb sets). Phase 2 (observer, reads KG): filter by character's knowledge (Confront requires KnowsDetails+ per [D-041](architecture.md#d-041-knowledge-graph-data-model)), apply POI priority flips, add contradiction markers. ~~Character-archetype verb variation implemented as Phase 2 observer filter rules (same crate: smuggler sees "Move/Stash", detective sees "Scan/Flag").~~ *[Removed Sprint 37 — archetype verb relabeling deleted; see supersession note above.]*
 - **Diegetic test:** Labels render on z-layer 6. If insert is off, labels disappear.