Add game-simulation package with multi-run balance testing, fix stalled-pipeline trap

Adds a full simulation harness (game-simulation package) with greedy/random strategies,
36-metric diagnostics, multi-run orchestration via child processes, and a statistical
interpreter. Includes 2.3x engine performance optimizations (research bonus caching,
per-DC dirty tracking, reduced allocations in tick pipeline, single-pass loops).

Fixes a critical balance bug where training pipelines stalled on insufficient VRAM would
permanently block training slots — the engine never re-checked stalled pipelines, and the
greedy strategy didn't pre-check VRAM requirements. This caused 20-25% of seeds to get
stuck in Scale-up era. All three fixes (engine un-stalling, strategy VRAM pre-check,
stalled pipeline cancellation) bring pass rate from 75% to 100% across 20 random seeds.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

packages/game-engine/src/systems/market/tamSystem.ts

@@ -92,13 +92,6 @@ function computeAttractiveness(
   return Math.max(0.01, score);
 }
 
-function softmaxShares(scores: number[]): number[] {
-  const maxScore = Math.max(...scores);
-  const exps = scores.map(s => Math.exp((s - maxScore) * SHARE_TEMPERATURE));
-  const sumExp = exps.reduce((a, b) => a + b, 0);
-  return exps.map(e => e / sumExp);
-}
-
 export function computeMarketShares(
   tam: TotalAddressableMarket,
   participants: ParticipantProfile[],
@@ -106,30 +99,48 @@ export function computeMarketShares(
 ): TotalAddressableMarket {
   const segments = { ...tam.segments };
   const segmentIds: TAMSegmentId[] = ['consumer', 'developer', 'enterprise', 'government'];
+  const n = participants.length;
+  const scores = new Array<number>(n);
+  const targetShares = new Array<number>(n);
 
   for (const segId of segmentIds) {
     const seg = segments[segId];
-    const scores = participants.map(p => computeAttractiveness(p, segId, qualityBaseline));
-    const targetShares = softmaxShares(scores);
+
+    for (let i = 0; i < n; i++) {
+      scores[i] = computeAttractiveness(participants[i], segId, qualityBaseline);
+    }
+
+    // Inline softmax
+    let maxScore = scores[0];
+    for (let i = 1; i < n; i++) { if (scores[i] > maxScore) maxScore = scores[i]; }
+    let sumExp = 0;
+    for (let i = 0; i < n; i++) {
+      targetShares[i] = Math.exp((scores[i] - maxScore) * SHARE_TEMPERATURE);
+      sumExp += targetShares[i];
+    }
+    for (let i = 0; i < n; i++) { targetShares[i] /= sumExp; }
 
     const oldShareMap = new Map<string, MarketShareEntry>();
     for (const entry of seg.shares) {
       oldShareMap.set(entry.playerId, entry);
     }
 
-    const newShares: MarketShareEntry[] = participants.map((p, i) => {
+    const newShares: MarketShareEntry[] = new Array(n);
+    let totalShare = 0;
+    for (let i = 0; i < n; i++) {
+      const p = participants[i];
       const old = oldShareMap.get(p.id);
       const oldShare = old?.sharePercent ?? 0;
       const migratedShare = oldShare + (targetShares[i] - oldShare) * SHARE_MIGRATION_SPEED;
-      return {
+      totalShare += migratedShare;
+      newShares[i] = {
         playerId: p.id,
         sharePercent: migratedShare,
-        customers: Math.floor(migratedShare * seg.totalSize),
+        customers: 0,
         attractivenessScore: scores[i],
       };
-    });
+    }
 
-    const totalShare = newShares.reduce((s, e) => s + e.sharePercent, 0);
     if (totalShare > 0) {
       for (const entry of newShares) {
         entry.sharePercent /= totalShare;
@@ -152,10 +163,7 @@ export function updateTAMGrowth(tam: TotalAddressableMarket, era: Era): TotalAdd
     const seg = segments[segId];
     const base = baseSizes[segId];
     const grown = seg.totalSize + seg.totalSize * TAM_GROWTH_PER_TICK;
-    segments[segId] = {
-      ...seg,
-      totalSize: Math.max(base, grown),
-    };
+    segments[segId] = { ...seg, totalSize: Math.max(base, grown) };
   }
 
   return { segments };