far shells: 3D cell bins for HasDownwardHit/HasUpwardHit candidates (L104, exact)

Near tiles legitimately need floor/ceiling boundary patches, but every hit query scanned the
tile's whole XY column. Direction-specific 3D cell bins conservatively index triangle bounds; the
existing predicates still decide every hit. Candidate visits per query -88..-99%; identical query
and hit counts. Hit-function cost ~1.7-1.9 -> ~0.35 s per FI/TN run. Near-ready @W12: FI 1.63 ->
1.52, TN 1.51 -> 1.38, VS 1.00 -> 0.92 s. All tile/whole hashes equal L103 (Clang/GCC W1/W8/W12).
The 24-28% "idle" is in the serial base-shape (~0.24 s) and far-parent (~0.13 s) phases; the
LOD0-1 drain is >= 99.4% busy. Fast tier PASS; Linux UE builds; FI Unreal arrival 1.96-1.98 s.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1

Standalone/Research/WorldAssemblerM3/Linux104/RunL104CensusExtraFI.py

@@ -0,0 +1,22 @@
+#!/usr/bin/env python3
+"""Add FI W12 L77 samples to tighten the final-source closure estimate."""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from RunL104Census import run
+
+ROOT = Path(__file__).resolve().parents[4]
+OUT = ROOT / "Saved/Linux/l104/census"
+
+
+def main() -> int:
+    rows = [run("FI", replicate) for replicate in range(4, 10)]
+    (OUT / "runs-extra-fi.json").write_text(json.dumps(rows, indent=2) + "\n",
+                                            encoding="utf-8")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())