jurisdiction-branch-map

Notes

• 2026-10-06 (open-items walkthrough, Dustin’s call): the 2026-08-11 JURISDICTION_BRANCH_MAP fix that had sat uncommitted in worktree heuristic-leakey for eight weeks landed on entirely-api master as e7557da and was deployed (all six containers rebuilt from /opt/unlikely-api at 08:58 ET, config.py md5 verified, pushed). Changes in config.py: “city of hagerstown” moves from manassas to baltimore (Dustin 8/11: Washington County is Baltimore territory, a city cannot route to a different branch than its county); “culpeper county” → richmond (only the town was mapped, the county fell through the Virginia state fallback to Manassas); “prince george county” replaces the dead “prince george, va” key (resolve_branch strips ”, va” before lookup and the leftover was swallowed by Manassas’s “prince george’s county” via the substring fallback); “city of fredericksburg” / “fredericksburg” / “fredericksburg city” get exact keys so the fallback no longer hits Manassas’s “city of frederick”; “richmond county” (Warsaw), “king and queen county” and “westmoreland county” get correct-spelling keys beside the SmartSuite-era misspellings. New tests/test_jurisdiction_branch_map.py (8 tests) guards every key resolving to its own branch and the two resolve_branch failure modes. resolve_branch is used by portal_reporting_service (branch roll-ups) and agent_cert_review (expected-branch check); existing rows in those jurisdictions are legacy imports with no branch code.