fix(simulation): PR #191 review round — n-clamp mirror, min_wl band quantization, coalescing coverage, fixture consumer

All seven Hoshe/Tyre findings addressed, none retracted:
- n-clamp/echo/staleness triangle (Tyre C1): client _clamp_window_n_mirror
  (bit-for-bit twin of the server clamp, canonicalize_district_center
  precedent) applied before _n is stored/sent; server test pins the
  quarter n=32 -> echo 16 contract.
- min_wl band quantization (Hoshe 1/Tyre C3): quantize_min_wl_m snaps to
  MIN_WL_BANDS_M {0, 32768, 16384, 8192, 4096} before cache key and echo
  (design doc §5's unbounded-key fix), reusing the one true
  OCTAVE_WAVELENGTHS_M array; docstrings now state the server-quantizes/
  client-sends-raw split; same-band cache-sharing test.
- coalescing granularity axis (Hoshe 2): two tests pin different-
  granularity requests as separate in-flight slots and same-granularity
  coalescing unchanged.
- orphaned fixture (Hoshe 3): test_protocol.gd consumer decodes
  atlas_response_ready_with_window.msgpack through the real IPC path and
  asserts the new fields.
- atlas_window_request coverage (Hoshe 4): new test file — stale-drop on
  granularity mismatch, old-server-shape defaults accepted, clamp mirror
  formula + wiring. First draft's quarter-via-request_now test would have
  passed for the wrong reason (request_now resets granularity by design
  until T-1153) — split into formula pin + reachable-path wiring proof.
- granularity type seam (Tyre C2): field + resolver docstrings state
  finer-only integer multiples with resolve_window_granularity as the
  single widening point; matching contract note added to the D-226
  T-1143-rulings amendment.

cargo --lib 1807/1807; goldens bit-identical; gdlint clean.

server/src/atlas/gen_queue.rs

@@ -1144,8 +1144,26 @@ mod tests {
     // -------------------------------------------------------------------
 
     /// Build a `DeriveWindow` work item pointing at a tiny test heightmap,
-    /// mirroring `analyze()`'s fixture shape.
+    /// mirroring `analyze()`'s fixture shape. `granularity` defaults to
+    /// district (matching every pre-T-1150 call site) via
+    /// `derive_window_at()` below — extended (PR #191 review, Hoshe 2) so
+    /// coalescing tests can exercise the granularity axis of
+    /// `window_supersede_key()` without a second near-duplicate helper.
     fn derive_window(body_id: &str, conn_id: ConnectionId, center: DistrictPos) -> GenWorkItem {
+        derive_window_at(
+            body_id,
+            conn_id,
+            center,
+            crate::atlas::layer_proxy::WINDOW_GRANULARITY_DISTRICT,
+        )
+    }
+
+    fn derive_window_at(
+        body_id: &str,
+        conn_id: ConnectionId,
+        center: DistrictPos,
+        granularity: u32,
+    ) -> GenWorkItem {
         GenWorkItem::DeriveWindow {
             body_id: body_id.to_string(),
             conn_id,
@@ -1161,7 +1179,7 @@ mod tests {
             }),
             center,
             n: 4,
-            granularity: crate::atlas::layer_proxy::WINDOW_GRANULARITY_DISTRICT,
+            granularity,
             min_wl_m: 0,
         }
     }
@@ -1282,6 +1300,82 @@ mod tests {
         );
     }
 
+    /// **PR #191 review, Hoshe 2 — zero coverage before this test.**
+    /// `window_supersede_key()`'s doc claims district and quarter requests
+    /// for the SAME `(connection, body)` are separate in-flight slots (the
+    /// key is `(conn_id, body_id, granularity)`, not `(conn_id, body_id)`).
+    /// Two submissions for the same connection+body but DIFFERENT
+    /// granularity must NOT coalesce — both survive as independent pending
+    /// items.
+    #[test]
+    fn submit_window_does_not_coalesce_different_granularity() {
+        let q = GenerationQueue::with_threads(1);
+        // See `submit_window_coalesces_same_connection_and_body`'s comment on
+        // why the occupier must be `analyze()`, not `FillChunk`.
+        q.submit(analyze("Occupier3"), GenPriority::Low);
+
+        let conn = ConnectionId(9);
+        q.submit_window(
+            derive_window_at(
+                "GranBody",
+                conn,
+                (0, 0),
+                crate::atlas::layer_proxy::WINDOW_GRANULARITY_DISTRICT,
+            ),
+            GenPriority::Immediate,
+        );
+        q.submit_window(
+            derive_window_at(
+                "GranBody",
+                conn,
+                (0, 0),
+                crate::atlas::layer_proxy::WINDOW_GRANULARITY_QUARTER,
+            ),
+            GenPriority::Immediate,
+        );
+        assert_eq!(
+            q.pending_count(),
+            2,
+            "same (connection, body) but DIFFERENT granularity must NOT coalesce — \
+             district and quarter are separate in-flight slots"
+        );
+    }
+
+    /// The coalescing-DOES-happen counterpart to the test above: two
+    /// submissions for the SAME `(connection, body, granularity)` still
+    /// collapse to one pending item — confirms the granularity axis didn't
+    /// accidentally loosen the existing same-key coalescing behavior.
+    #[test]
+    fn submit_window_coalesces_same_connection_body_and_granularity() {
+        let q = GenerationQueue::with_threads(1);
+        q.submit(analyze("Occupier4"), GenPriority::Low);
+
+        let conn = ConnectionId(11);
+        q.submit_window(
+            derive_window_at(
+                "SameGranBody",
+                conn,
+                (0, 0),
+                crate::atlas::layer_proxy::WINDOW_GRANULARITY_QUARTER,
+            ),
+            GenPriority::Immediate,
+        );
+        q.submit_window(
+            derive_window_at(
+                "SameGranBody",
+                conn,
+                (5, 5),
+                crate::atlas::layer_proxy::WINDOW_GRANULARITY_QUARTER,
+            ),
+            GenPriority::Immediate,
+        );
+        assert_eq!(
+            q.pending_count(),
+            1,
+            "same (connection, body, granularity) must still coalesce to one pending item"
+        );
+    }
+
     // -------------------------------------------------------------------
     // TerrainAnalysisCache (T-1137, PR #187 review — Tyre C1)
     // -------------------------------------------------------------------