Capture current mission, AI, map and administration development

Include the current addon, terrain integrations, tools, regression tests and recorded development evidence.

docs/mortar-rapid-fire.md

@@ -1,6 +1,6 @@
 # Rapid mortar barrage — 27 September 2026
 
-An authorized HQ fire mission retains eight HE rounds distributed over an 80-metre radius. Mission-owned AI mortar crews use a dedicated behavior tree with the two fixed 1.5-second firing-loop delays removed. The native loading animation, physical projectile event, ammunition transfer, per-round ballistic solve and spread still determine the real firing rate. The 0.75-second settling pause after leaving the aiming seat is retained.
+An authorized HQ fire mission retains eight HE rounds. Unobserved fire distributes them over an 80-metre radius; the observed-fire correction described below tightens subsequent fire. Mission-owned AI mortar crews use a dedicated behavior tree with the two fixed 1.5-second firing-loop delays removed. The native loading animation, physical projectile event, ammunition transfer, per-round ballistic solve and spread still determine the real firing rate. The 0.75-second settling pause after leaving the aiming seat is retained.
 
 The tree is derived from the installed Arma Reforger 1.8.0.13 `AI/BehaviorTrees/Chimera/Soldier/StaticArtilleryBehavior.bt`. Only its resource identity and the two delay values differ. It is selected only for a mortar registered with the mission mortar service; unrelated artillery keeps its stock tree. Player firing actions are unchanged.
 
@@ -19,3 +19,17 @@ Final native run passed all 57 assertions with eight distinct physical projectil
 The runner recorded `smoke-completed`, source/dependency hashes unchanged, and cleanup confirmed. The generic report still returns nonzero for the same 13 installed dependency/prefab errors seen in the stock baseline; it reports no script/runtime failures or new engine errors. All repository checks passed, including the executable last-shot regression. Removing just the completion wait makes that regression fail.
 
 These results qualify local native Workbench launch behavior and ammunition accounting. Multiplayer replication, actual impact positions and damage are not measured by this fixture. Evidence: `.local/analysis/mortar-cadence-2026-09-27/`.
+
+## Observed fire correction (2026-10-11)
+
+Mission-owned AI projectiles are tracked by identity until the engine's actual explosion callback. A registered enemy soldier may spot the splash if their own native perception currently sees the target, they are within 600 m of the splash, the splash is within 200 m of that sighting, and a separate unobstructed ray confirms a view of the splash in the same viewing sector. Guards, support crews and mounted soldiers are eligible; a designated observer class is not required. Missing explosions, target memory alone and predicted impact positions cannot authorize corrections.
+
+The report records approximate left/right and short/long error in five-metre increments relative to the battery's firing direction. It travels through the existing squad radio uplink and an HQ-to-battery downlink. Broken routes discard it. After delivery, the next new native aim uses the latest delivered target observation and an eight-metre spread disk, with a horizontal ballistic calibration capped at 60 m. The calibration compares the commanded point with the actual splash, so intentional spread and player movement are not counted as systematic aiming error. Earlier calibration is not accumulated twice. Older shells landing later cannot overwrite a newer fired shell's correction.
+
+Splash sightings enter the ordinary HQ observation history only after their uplink arrives. The correction then carries a coherent fire-support snapshot (position, observation time and estimated velocity) through its downlink. This avoids replacing moving-target intelligence with a position-only report from an ordinary soldier. Existing prediction leads an observed moving target by report age plus shell flight/setup time; conflicting or insufficient movement history leaves velocity unknown rather than reading the player’s live transform.
+
+An existing barrel command or committed shell is unchanged. Revalidation preserves that command's original offset. Fresh target reports, finite ammunition, original salvo limits, settlement/friendly exclusion and airspace coordination still gate every round. Calibration expires 45 seconds after the observation and does not follow a different target or a gun moved more than two metres.
+
+Work is bounded: 64 tracked shells, 32 correction reports, one splash survey per HQ tick, at most 32 groups and eight splash rays per survey, and a five-second splash observation window. The journal distinguishes `SPLASH`, `MORTAR SPOT`, `SPOT RECEIVED`, `SPOT ADOPTED`, `SPOT LOST` and `SPOT UNOBSERVED`. This implementation routes correction data; it adds no new synthesized voice line.
+
+Validation: `test_mortar_spotting.py` executes the production correction, delivery, selection and adoption methods against API doubles, covering sign, repeated calibration, out-of-order impacts, stale/failed delivery, moved guns, changed targets, existing aim/commit protection and bounded cleanup. Existing mortar policy regressions pass. Hidden Reforger 1.8.0.13 Workbench compilation passed (`err-mortar-motion-20261011-b`, Game CRC `13c3e0fb`). Actual explosion-to-observer-to-adjusted-impact behavior still needs a live mission check; the earlier launch-cadence evidence does not qualify this new impact feedback.