Validate independent packed addons and complete physical mortar salvos

docs/validation.md

@@ -331,3 +331,33 @@ For the six newly checked 091306/091641/092237/092512/092923/game093115 archives
 [102737 ordinary companion startup](evidence/game-2026-09-19-102737/report.json) runs the regular executable with all five successor addons and no Workbench defines or smoke fixture. The expected Chat header/MissionSystems initializes, preparation reaches PRISON, and the real sole local controller receives a living canonical prisoner. Fifteen seconds of stable readiness, zero captured errors, unchanged addon source bytes and owned cleanup are verified. The scoped diagnostic noSplash workaround and standalone round-local identity remain explicit; the separately archived vanilla control reproduces the initial splash error without this mod. No normal visible splash, input, full ordinary escape or multiplayer acceptance is inferred.
 
 [103025 checkpoint companion regression](evidence/native-2026-09-19-103025/report.json) repeats two complete companion-loaded escape/restart rounds, 96 native assertions and zero captured errors on the exact source shared with casualty regression 102037 and ordinary Chat startup 102737. Teardown diagnostics remain disabled, and the opt-in failed flight experiment is inactive. This source qualifies the development clean-runtime checkpoint; all seventeen release acceptance cases remain pending. A subsequent experimental actor-local flight-input branch must invalidate current-source qualification until separately tested.
+
+[104048 actor-local flight refusal](evidence/native-2026-09-19-104048-flight-action-refusal/report.json) compiles the optional upstream action path and proves the supplied manager is the exact pilot agent manager, distinct from the global manager, with HelicopterContext active. It fails before engine start with zero virtual writes. [104256 context diagnostics](evidence/native-2026-09-19-104256-flight-context-refusal/report.json) then records four callbacks where ActivateContext returns false while IsContextActive is true, with valid timestep/input/simulation. The fixture incorrectly required both activation return and active state; no meaning for that return was documented. The correction queries existing active state first, activates only when inactive, and requires observed active state before any writes. Native action consumption and physical flight remain separate mandatory gates. Both failures remain retained.
+
+[104438 consumed-input takeoff](evidence/native-2026-09-19-104438-flight-drift-refusal/report.json) proves the corrected actor-local action path reaches native collective/cyclic inputs and existing rotor signals. The mandatory0.05 collective gate agrees on distinct ticks; the same aircraft takes off using real thrust, reaches5.318m above its settled origin and consumes finite fuel. It then fails the unchanged8m horizontal-drift bound before stable hover or landing. This is meaningful native input/takeoff evidence, not a flight-controller acceptance or production HQ aircraft integration. Further work requires motion telemetry and controlled stabilization under the same physical bounds.
+
+[105226 flight motion diagnostic](evidence/native-2026-09-19-105226-flight-motion-refusal/report.json) repeats real input consumption and takeoff with read-only velocity, acceleration, angular-rate, basis and native rotor-axis observations. The same airframe reaches6.8385m and fails at8.00134m horizontal displacement; lateral velocity has already begun reversing. The native rotor-axis API is documented in local space but is not itself a measured thrust direction. No hardcoded rotor trim or safety-bound increase follows from this trace. Spectator guidance compiles in this Core run; ordinary hint rendering and input remain unqualified.
+
+[104634 first private native packaging attempt](evidence/package-2026-09-19-104634/package-report.json) copies validated authored Core source into a fresh private session. Workbench prints Packaging project successful and naturally returns0, producing data.pak, addon.gproj and resourceDatabase.rdb. However, native shutdown reports UI resource leaks and Gamescope remains alive until the original runner's300s deadline. Cleanup is confirmed and both captured repository/copied authored bytes remain unchanged; the attempt is a failure, not a release build. Independent read-only inventory finds107 packed entries byte-identical to all107 captured Core source files after metadata/project exclusion; this validates content correspondence, not packed-only engine loading or the noisy build.
+
+[105441 full-salvo refusal](evidence/native-2026-09-19-105441-salvo-expiry-refusal/report.json) compiles the complete eight-round observer and preserves the original27 physical HE shells. A single explicit inert-target observation travels through real uplink, HQ and downlink; the same battery commits and launches two distinct physical projectiles. Its immutable report then reaches60s age and firing correctly closes without a third launch. The strict full-salvo case fails, while the refusal remains consistent with stale-report policy. Audit confirms fresh later HQ contacts currently have no downlink path into an already active battery; implementing an explicit fresh-report update path is separate from weakening timestamps or inventing observations.
+
+[105709 flight damping experiment](evidence/native-2026-09-19-105709-flight-damping-refusal/report.json) increases only velocity damping0.3→0.6. It again consumes actual controls and fuel, and reaches roughly9m altitude. It later oscillates horizontally and exceeds the unchanged flight envelope; no stable5s hover or landing passes. Higher outer-loop damping alone is insufficient with the observed native attitude response. This remains opt-in WORKBENCH investigation, never a production aircraft controller.
+
+[105949 silent packaging attempt](evidence/package-2026-09-19-105949-silent/package-report.json) repeats native Core packing with documented-wbSilent, retaining actual successful packing messages and natural native exit0. The corrected runner now distinguishes native completion from its remaining owned display wrapper and confirms cleanup without a timeout. Native UI resource-leak errors still occur, so the attempt remains failed; silent initialization does not solve that environment issue. All captured source bytes remain unchanged.
+
+[110404 empty-addon packaging control](evidence/package-control-2026-09-19-110404/package-report.json) contains only the five owned project descriptors, with no successor scripts, worlds, compositions or UI resources. The same installed native packer naturally exits0 and repeats the UI resource-leak shutdown errors. Its private input repository remains unchanged and cleanup is confirmed. This isolates the shutdown symptom from successor payload content in this environment; it neither clears the earlier errors nor qualifies a build, mission, package consumption or release.
+
+[110127 first packed-only startup](evidence/packed-core-2026-09-19-110127-path-refusal/report.json) mounts only the expected packed successor/vanilla inputs with no native errors, but the runner rejects the engine's exact `$profile:../addons/` search alias and stops before readiness. The corrected verifier resolves only that literal alias to the captured private profile's addon directory and rejects traversal/other aliases; the failed run remains latched in its archive.
+
+[110729 packed-only ordinary Core startup](evidence/packed-core-2026-09-19-110729/report.json) passes after the verifier correction. The regular executable loads exactly engine Core, vanilla ArmaReforger and packed successor Core, with exact package/database paths, normal non-Workbench modules and no loose successor or original addon. The actual Core scenario/MissionSystems reaches PRISON, equips the real local canonical prisoner, and remains ready for15s with zero captured errors. All3runtimeartifact bytes and sealed console hashes are preserved in the independent archive; ownedcleanup isconfirmed. This proves consumption of that specific native package despite its separately retained packaging shutdown errors. It does not qualify current later source, manual gameplay, full escape, multiplayer or release.
+
+## 19 September — packaged closure and mortar completion
+
+The [five-addon packaging archive](native-companion-packaging.md) retains all authored inputs and fifteen runtime artifacts. All147 PAK entries match captured source, but all five native packer invocations retain the same UI shutdown errors as the empty control and remain unqualified. Packed Core ordinary startup110729 is a separate successful artifact-bound check.
+
+The fresh-report salvo initially fired all eight rounds but failed natural completion at 111915. Native activity terminal events can be followed by a reset polled state; the exact-activity/order event latch now preserves the terminal event and detaches before cancellation cleanup. The [112900 rerun](evidence/native-2026-09-19-112900-mortar-refresh/report.json) passes 55 assertions with zero captured errors, three delivered refreshes, eight actual launches, nineteen original shells remaining and 25 seconds without replay. The expiry classification failure110955 and immutable-report expiry105441 remain retained.
+
+Powered flight112320 refused before engine start because native autohover also changed collective. Manual angular damping112502 held horizontal drift below 8 metres but still exceeded the unchanged 18 metre height limit. Both failures are retained; hover/landing and operational HQ aircraft remain unqualified.
+
+The [113315 packed Chat ordinary startup](evidence/packed-chat-2026-09-19-113315/report.json) succeeds with all five 111731 PAKs, exact seven loaded addons including engine/vanilla, fifteen unchanged artifact files, canonical prisoner and 15 seconds ready; zero captured errors and owned cleanup confirmed. Its separate packed provenance does not qualify later loose-source edits or waive packaging shutdown errors.