Add civilian traffic variants, farm tractors and UI scaling; record Workshop 0.0.143

docs/evidence/review-1.9-2026-11-06/README.md

@@ -0,0 +1,74 @@
+# Reforger 1.9 mission review
+
+Requested: apply civilian vehicle variety (3), check ACE repacking (1), AI
+spawning/dormancy (4), and UI scaling (5). Tested with installed 1.9.0.37.
+
+## Changes
+
+- Moving civilian traffic discovers ordinary S105/V2101 variants once, with
+  80% car / 20% existing S1203 van selection. The native catalog contained 19
+  moving-car entries. Rally cars and tractors remain outside moving traffic.
+- T601 stays in the parked pool. Two native road locations left its original
+  AI driver seated without starting the engine; the diagnostic recorded engine
+  off, handbrake on and negligible movement. V2101 drove on those tests. No
+  speculative engine or vanilla prefab override was added.
+- Optional Rally Z7211 civilian tractors can appear beside supported barns and
+  cowsheds. Farm selection is capped at 15%, shares the existing parked fleet
+  limit, and uses wider ground/collision and taller visibility checks. Existing
+  claim and distant cleanup rules remain in use. No new DLC dependency.
+- The admin panel shrinks only when necessary to fit the viewport, without
+  changing the user's saved UI scale. Baseline 150% scale clipped vertically;
+  final native measurements fit at 75%, 100%, 125% and 150%.
+
+## Checks without changes
+
+Installed ACE Magazine Repack 1.4.3 already handles inventory drag/drop through
+its `SCR_InventoryMenuUI.MoveItemToStorageSlot` override and server-side ammo
+transfer. It checks matching magazine wells and available capacity. This is a
+source inspection, not a new multiplayer ACE/RHS interoperability test. No
+duplicate repacking implementation was added.
+
+Site squads, patrols and related groups already request queued native spawning.
+Town garrisons instead create one actor per background callback, with mission
+placement and exclusion checks. Convoys and aircraft have explicit AI activity
+requirements. The isolated native world reported partial fill and dormancy
+enabled with an active AI limit of 128; its zero active count is not a production
+load measurement. No priority or global dormancy change was made. Native group
+despawn/recreation can reset inventory and health, so further optimization needs
+to preserve the mission's finite ammunition, casualties and crew continuity.
+
+Official context: [1.9 changelog](https://reforger.armaplatform.com/news/changelog-november-4-2026),
+[September modding update](https://reforger.armaplatform.com/news/modding-update-sept-24-2026),
+[October modding update](https://reforger.armaplatform.com/news/modding-update-oct-21-2026).
+
+## Evidence and limits
+
+`final.log` records both original S105/V2101 drivers travelling over 100 metres,
+native tractor placement through production Plan/Pending ground, visibility and
+collision checks, and UI geometry. The fixture supplies a stationary observer
+and a spawned native barn; it forces that barn's selection to exercise placement.
+It does not test random farm frequency or multiplayer observers. UI measurements
+use the 501x225 Workbench game viewport, not manual controller/TV usability.
+
+`runner.json` reports `smoke-completed`, unchanged source/external inputs and
+confirmed cleanup. `production-source-comparison.json` verifies all changed
+production scripts/layout match the tested snapshot. The test-only probe and
+launch manifests are included. Tests used hidden, muted Gamescope and a disposable
+Proton prefix; the normal mission startup was suppressed in the fixture.
+
+The strict runner still exits nonzero because the complete log contains resource
+errors: external localization references, legacy vehicle prefab fields, a DLC
+ownership request failure, and fixture path-only resource lookups. There are no
+Game script compilation errors or VM exceptions in the final run. Tractor resource
+and seat checks passed, but live client DLC entitlement/driver access is unqualified.
+This is not a clean-all-engine-logs or dedicated multiplayer certification.
+
+`repository-check.log` records the full passing repository validator. After the
+final T601 exclusion, catalog rules, traffic, garage lifecycle, panel fit and menu
+source tests passed again. `ui-baseline.log` and `t601-diagnostic.log` preserve the
+earlier failures that informed the changes. Other exploratory runs included a
+test-only API typo and observer-fixture failure; neither changed production logic.
+
+Workbench subsequently confirmed successful upload of 0.0.143 at 12:35:36 on
+2026-11-06. See `publication.json`; the changelog version was recorded with
+`tools/workshop_changelog.py --published` before committing this release.