docs: JTS #1093 Ozaki orientation filter lane (GREEN)

[grootstebozewolf]

Summary

• Complete the orientation lane for locationtech/jts#1093 (Ozaki et al. production orientation filter).
• Expand the scout to a three-way Stage A mirror: JTS-master (DP_SAFE_EPSILON = 1e-15) vs Ozaki (#1093) vs corpus Shewchuk Stage A (ccwerrboundA), gated on ORIENT_EXACT.
• GATE GREEN on 45 vectors: 0 CERTAIN vs exact conflicts for all three filters; Ozaki CERTAIN while master UNCERTAIN on 4 same-sign band cases; Ozaki ↔ corpus Shewchuk certainty splits 0.

Files

Path	Role
oracle/gen_jts1093_orient_scout.py	Three-way Stage A mirror + oracle/FFI gate
oracle/jts1093_orient_vectors.txt	45-vector table
docs/jts-1093-orient-lane-2026-08.md	Session write-up
tests/Discussion839Mre/jts-1093-comment.md	Draft comment for jts#1093
docs/jts-open-prs-scout-2026-08.md / docs/ffi-rungs-2026-08.md	Status → GREEN

Gate (local WSL)

Vectors: 45
EXPECTED pin failures: 0
Ozaki / Shewchuk / JTS-master CERTAIN vs exact conflicts: 0 / 0 / 0
Ozaki tighter than JTS-master: 4
GATE: GREEN

Test plan

• python oracle/gen_jts1093_orient_scout.py --wsl-oracle …/oracle_bin --wsl-ffi-probe …/ffi_probe (exit 0)
• Optional: post tests/Discussion839Mre/jts-1093-comment.md on Change Shewchuk Orientation filter to Ozaki et al. filter. locationtech/jts#1093
• Optional: wire make -C oracle jts1093-orient-scout into CI smoke later