[AZ-808] [AZ-809] [AZ-810] [AZ-811] [AZ-812] Cycle 8 perf run

8/8 scenarios PASS within threshold. Cycle-8 strict-validation
overhead is below percentile resolution on every measured
endpoint.

PT-06 (route creation) required one in-cycle perf-script fix:
add requestMaps=false + createTilesZip=false to the body to
satisfy AZ-809's no-defaulting rule. The script had already
been updated for AZ-812's wire rename during cycle 8 but missed
AZ-809's newly required fields. Production code is correct; only
the perf probe was stale.

Report: _docs/06_metrics/perf_2026-05-23_cycle8.md. Trend vs
cycle 7 is flat within noise band on every scenario.

Known harness quirks (pre-existing, not cycle-8 regressions)
surfaced and documented for cleanup:
- PT-07 cross-run cache pollution (hard-coded base coords)
- PT-01 "cold" misnomer (tile cached on disk since cycle 5)
- PT-03 cached-by-PT-02 side effect (cycle-7 note carried forward)

Auto-chains to Step 16 (Deploy).

Co-authored-by: Cursor <cursoragent@cursor.com>

_docs/_autodev_state.md

@@ -2,13 +2,13 @@
 
 ## Current Step
 flow: existing-code
-step: 14
-name: Security Audit
+step: 15
+name: Performance Test
 status: completed
 sub_step:
   phase: 5
   name: report-rendered
-  detail: "verdict: PASS_WITH_WARNINGS (1 Medium F-AZ809-1 + 2 new Lows + 3 carry-over Lows + 1 carry-over Medium D2-cy4 test-runtime only)"
+  detail: "verdict: PASS (8/8 scenarios). Required one in-cycle AZ-809 contract-sync fix to PT-06 body. Report: _docs/06_metrics/perf_2026-05-23_cycle8.md."
 retry_count: 0
 cycle: 8
 tracker: jira