wgpu streaming: screen-space AABB priority + grace period + interactive heartbeat

The chunk priority metric is now the 2D projected pixel area of the
chunk's AABB on screen — 8 corners projected through view-projection,
2D axis-aligned bbox of the projected points, clamped to viewport.
This replaces the prior bounding-sphere-radius² metric, which was a
3D approximation: it treated a 322 × 55 × 5 m slab as a 163 m sphere,
giving it the same huge priority face-on or edge-on. The new metric
genuinely answers "what would this chunk's AABB cover if rendered
solid given the current camera and viewport."

Newly-loaded chunks get a 30-frame grace period at full priority
(visibility_history floor temporarily forced to 1.0). Without it,
just-loaded chunks crashed to history=0 → effective priority = pri ×
0.05 → immediately reverse-swapped by the chunk they displaced.
Cycle starved the per-frame load budget so candidates ranked below
the cyclers never got attempted. 30 frames = HISTORY_ALPHA's time
constant — enough for visibility_history to develop meaningfully.

EVICT_PRIORITY_RATIO bumped 1.21 → 2.0 to suppress more swap noise
between similar-priority chunks.

Interactive heartbeat log added: every render in non-bench mode prints
[frame] with fps, ms, obj, sub_draws, hiz_rej, cull, stream, chunks
breakdown (resident/frustum/total + missing count), VRAM, model count.
Every 30 frames when something's missing, also dumps:
- top 8 models by missing chunk count
- top 20 missing chunks by priority (with AABBs)
- bottom 5 residents by effective priority
- all chunks of brace.ifc (one-off diagnostic, hardcoded
  for the brace-visibility investigation)

The heartbeat made the streaming bug visible: a brace model that
isolation-loads correctly is missing in the full set because slabs
covering more pixels win the priority contest. Per-model fairness or
manual pinning are the remaining options if pixel-area + grace +
hysteresis isn't enough — left for follow-up so the user can decide
based on real testing.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

src/ifcviewer-wgpu/WgpuModelGpuData.h

@@ -178,6 +178,28 @@ struct WgpuModelGpuData {
         // total_visible_draws > 0). The evictor picks the smallest value
         // among non-visible resident chunks when it needs to free VRAM.
         uint64_t last_visible_frame_idx       = 0;
+        // EMA-smoothed visibility score, in [0, 1]. Bumped each frame
+        // toward 1 when total_visible_draws > 0 (the chunk's instances
+        // passed frustum + contribution + HiZ), toward 0 otherwise.
+        // Time constant ~30 frames. Used by the streaming evictor to
+        // de-prioritise chunks that are technically in the frustum but
+        // consistently HiZ-occluded — e.g. interior pipes behind a
+        // building's exterior walls. The smoothing prevents thrash from
+        // momentary HiZ flicker (a wall briefly visible behind a panning
+        // window doesn't displace the window from the pool).
+        float    visibility_history           = 0.0f;
+        // streaming_frame_idx_ when this chunk was last loaded. The
+        // evictor grants newly-loaded chunks ~30 frames of grace at
+        // full priority (max history factor = 1.0) so they have time
+        // for visibility_history to develop. Without this, a just-
+        // loaded chunk's effective priority drops to contribution ×
+        // 0.05 next frame, and the chunk it displaced — back as a
+        // candidate at full priority — re-displaces it: infinite
+        // cycle between equal-priority chunks. The cycle prevents any
+        // lower-priority candidate (e.g. a structural-brace chunk
+        // ranked position 20 in the missing list) from ever getting
+        // attempted.
+        uint64_t loaded_frame_idx             = 0;
     };
     std::vector<Chunk> chunks;