Fix Flagship visuals, beam aiming, results and Frame controls

docs/DEVELOPMENT.md

@@ -29,13 +29,13 @@ After installing the documented Python requirements, run from the repository roo
 python -m unittest discover -s tools -p "test_*.py"
 ```
 
-The current suite passed **44 Python tests**, including empty native drone-list, portable-launcher and modal-HUD hook/preflight regressions. The tests exercise synthetic archive/layout/font extraction, bridge schema/input translation, native-pixel filtering, raw frame publication and launcher preflight/error handling. They do not launch FTL. Temporary fixture files are created by the tests. Final Godot and fresh native-game checks for the latest targeting/jump/drone batch are pending; earlier results and focused drone fixture checks are recorded in [STATUS](STATUS.md). Physical headset coverage remains separate.
+The current correction passed **46 Python tests**, including empty native drone-list, portable-launcher, modal-HUD hook/preflight and native result-window regressions. The tests exercise synthetic archive/layout/font extraction, bridge schema/input translation, native-pixel filtering, raw frame publication and launcher preflight/error handling. They do not launch FTL. Temporary fixture files are created by the tests. Native beam placement/queued-shot acceptance and defeat/main-menu return pass; live native beam/miss outcomes remain unverified. Production state is restored with preserved campaign hashes; development desktop launch/exit and SteamVR preflight pass. Both standalone wrapper checks and source-only package checks pass; see the current delivery evidence in STATUS. Recorded evidence is in [STATUS](STATUS.md). Physical headset coverage remains separate.
 
 ## Godot checks
 
 Use a **separate development checkout or disposable local-data folder** with the required extracted assets. Some suites write screenshots/test frames under `local_game_data/`; do not run them over an active production bridge or package those outputs. Create the local output folder before graphical suites if it is missing.
 
-The commands below list eleven suites, including the new target-lock and drone resource/variant checks. The preceding HUD/wheel/bar revision passed nine suites; the following clearance/menu correction reran focused Vulkan HUD-layout and input checks plus native before/after capture checks. The latest batch's final combined Godot run and fresh native proof are pending. See [STATUS](STATUS.md) for evidence and limits. A fresh source checkout lacks game assets; model/input/font checks may fail until local extraction is complete. Existing installations should also rerun [asset extraction](INSTALLATION.md#5-extract-local-presentation-assets) to import the eight placed reticle PNGs.
+The commands below list twelve suites. The current correction passed eight headless suites and four graphical Vulkan suites, including the new ship-status render check. See [STATUS](STATUS.md) for evidence and remaining native/delivery limits. A fresh source checkout lacks game assets; model/input/font checks may fail until local extraction is complete. Existing installations should also rerun [asset extraction](INSTALLATION.md#5-extract-local-presentation-assets) to import the eight placed reticle PNGs, and refresh [native hooks](INSTALLATION.md#4-resolve-the-local-executable-hooks).
 
 ```powershell
 godot --headless --xr-mode off --path . --script tools/test_combat.gd -- --demo
@@ -47,14 +47,17 @@ godot --headless --xr-mode off --path . --script tools/test_environment.gd -- --
 godot --xr-mode off --path . --script tools/test_ui.gd -- --desktop
 godot --xr-mode off --path . --script tools/test_controller_ui.gd -- --demo --desktop
 godot --xr-mode off --path . --script tools/test_frame_transport.gd -- --desktop
+godot --xr-mode off --path . --script tools/test_ship_status_render.gd
 godot --headless --xr-mode off --path . --script tools/test_hud_layout.gd -- --desktop
 godot --headless --xr-mode off --path . --script tools/test_damage_visuals.gd -- --demo --desktop
 ```
 
-The UI/controller/transport suites use actual GPU rendering, so keep their graphical mode. Headless runs of them are not equivalent verification. These suites cover effects, room/beam picking, controller context/bindings, model/task distinctions, environmental presentation, original local fonts, UI priority/crops and the raw frame protocol. HUD-layout checks cover headset-relative placement, swept collision prevention and transformed ship/shield clearance. Controller checks include native empty/malformed equipment collections. Damage-visual checks cover permission-aware room colors, absence of floor status bars, native miss presentation and oxygen-dependent breach effects.
+The UI/controller/transport/ship-status suites use actual GPU rendering, so keep their graphical mode. Headless runs of them are not equivalent verification. The ship-status suite needs owner-local extracted ship assets; its default state is a labelled synthetic renderer fixture using `rebel_long`. To inspect a recorded native enemy instead, add `-- --state=res://local_game_data/recorded-enemy.json --output=res://local_game_data/ship-status`. The corresponding layout/art must also be extracted. Neither mode drives FTL. These suites cover effects, room/beam picking, controller context/bindings, model/task distinctions, environmental presentation, original local fonts, UI priority/crops and the raw frame protocol. HUD-layout checks cover headset-relative placement, swept collision prevention and transformed ship/shield clearance. Controller checks include native empty/malformed equipment collections. Damage-visual checks cover permission-aware room colors, absence of floor status bars, native miss presentation and oxygen-dependent breach effects.
 
 Target-lock checks cover numbered player-owned native targets, autofire color, beam endpoint direction, pinpoint beams, flak radius, transformed/pause attachment, receiver visibility and cleanup. Drone checks cover distinct Mk II hardware, shared immutable body/exhaust/emitter geometry and independent emission state. Combat checks include native drone muzzle/render-space placement, fixed bullet launch points, continuing beams and authoritative progress/pause/outcome handling. Jump checks use native `jumping` and arrival-dialog fixtures; the action-wheel context checks include resetting each opening to weapons/drones.
 
+For the current correction, cover one-HP hull cells for differing enemy hull maxima; separate artillery mounts and native shot origins; shield shutdown/charge and native cloak restoration; forward-only miss progress; free beam deck/gap endpoints under encounter transforms; complete result-screen cleanup; 5 cm HUD clearance; and the right Pause/Menu versus left View actions. Native checks must separately establish real artillery/shield/cloak state, accepted beam endpoints, result-window detection and supplemental target isolation. Rendered fixtures establish presentation behavior and must be labelled as fixtures.
+
 No test command above drives a native game or modifies its saves. Separate **live bridge checks** do run the game and can advance a run or change equipment/state.
 
 ## Native integration checks
@@ -72,6 +75,8 @@ When adding a system or command, compare the visible native outcome and snapshot
 
 A preceding native missed-event check used a private projectile and forced `Evasion.MISS` to exercise the actual flag/event path. It is integration evidence, not a measurement of random evasion probabilities. Enemy room condition colors respect the native visibility rules; floor health/status bars are no longer rendered. Do not infer permission to reveal other hidden state from public icon color or crew-room fog alone.
 
+For the October 5 correction, native beam aim/release and its queued-shot transition are verified. The tested newly equipped and original saved missile/burst factories queued shots without releasing live projectiles, so this session does not establish a native beam sweep or missed-shot outcome. Keep that limit explicit and do not assign a speculative cause. Graphical miss fixtures prove forward presentation/cleanup only. Controlled Victory/defeat checks prove native result-screen integration, not a completed campaign.
+
 The preparer, resolver and launcher reject unknown executable fingerprints. To support another executable, identify a known owned build, implement and validate a complete matching route and keep the rejection path. Do not disable fingerprint checks or reuse addresses from a different build.
 
 ## Profiling and visual inspection