wip(streaming): priority by class and level -- near ring 53.6s -> 1.2s, but crossing test fails

Scheduling rule: CollisionCritical -> Edited -> LOD0 -> Horizon, with coarse work capped so 4 task slots stay reserved for near work; result queues follow the same order; epoch and cancellation handling unchanged.

Measured at MaxClipLevel=5 / RenderDistanceChunks=768: spawn ring request-to-ready 53.632 s -> 1.171 s; spawn ring from trace start 68.252 -> 62.549 s; moving first complete ring 66.808 -> 63.917 s, warm ring p50 0.267 s, per-tile moving LOD0 p50 69.4 -> 62.0 ms; full settle of 10,841 tiles 110.7 -> 106.9 s. Gate stress PASS (139.99 m, 0 full stops); static perf PASS.

⚠️ NOT SAFE TO MERGE YET: the strate-crossing test fails in both patched runs with capsule_overlaps_pending_floor -- the exact guarantee the gate exists for. Codex suspects the owner's current DA_Settings (seal 50, clip level 5) rather than this change, since older passing captures used different settings. Control needed: same test, same assets, on 370c39e.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MmSoF64BbRFmk9C3p1VsBJ

Source/VoxelForge/Public/VoxelSettings.h

@@ -53,6 +53,13 @@ public:
 	UPROPERTY(EditAnywhere, BlueprintReadWrite, Category = "Voxel|Streaming", meta = (ClampMin = "0"))
 	int32 MaxGenerationWorkerCores = 0;
 
+	// Keep this many task slots available for collision-critical, edited, and level-0 terrain while
+	// coarse horizon work is in flight. A positive reserve prevents a full far-field queue from
+	// delaying a new near tile because running tasks cannot be preempted. The runtime clamps the
+	// value below the effective task budget; 4 is the conservative default for the ship settings.
+	UPROPERTY(EditAnywhere, BlueprintReadWrite, Category = "Voxel|Streaming", meta = (ClampMin = "0", ClampMax = "16"))
+	int32 NearTaskReserve = 4;
+
 	// Nombre max de meshes appliqués (upload GPU) par frame. Seuls les vrais applies
 	// comptent (chunks vides/périmés se vident gratuitement). À 32³ chaque apply est léger
 	// (~0.1 ms) — tunable en live sur l'asset : montez (8-16) si le remplissage traîne,