feat: PolicyControllerAi integration — Forge AI decision interception via policy server

## New File
- PolicyControllerAi.java (840 lines): Extends Forge's PlayerControllerAi
  to intercept AI decisions at chooseSpellAbilityToPlay() and route them
  through the Python policy server. Falls back to Forge built-in AI on
  any failure. Inner PolicyAiController overrides AiController with full
  legal action gathering via ComputerUtilAbility.

## Wiring
- GameSession.java: Import + Javadoc for PolicyControllerAi decision flow.
  Logs decision pipeline when policy server connects. End-of-game reward
  submission via PolicyControllerAi lifecycle.
- BatchRunner.java: New setPolicyServer(url) method. Appends
  --policy midrange --policyServer <url> to Forge subprocess command when
  policy server URL is configured.

## Path Updates (Forge-Repo rename)
- README.md: Forge link → KoalaTrapLord/Forge-Repo, build path updated
- .env.example: FORGE_JAR_PATH and FORGE_WORKING_DIR → Forge-Repo
- lab_api.py: Auto-detect paths try Forge-Repo first, forge-repo as legacy

## MacroActionExecutor
- Added public isRemovalCard(), isDrawCard(), isRampCard() wrappers so
  PolicyControllerAi can consult the same card-role databases.

Refs: Issue #83 Phase 2.4

src/main/java/commanderailab/batch/BatchRunner.java

@@ -1,6 +1,7 @@
 package commanderailab.batch;
 
 import commanderailab.ai.AiPolicy;
+import commanderailab.ai.PolicyControllerAi;
 import commanderailab.ml.DecisionExtractor;
 import commanderailab.ml.DecisionLogger;
 import commanderailab.schema.BatchResult;
@@ -76,6 +77,12 @@ public class BatchRunner {
     private boolean useSimplifiedAi = false;   // Use faster AI profile for sims
     private int aiThinkTimeMs = -1;            // -1 = Forge default; >0 = cap AI think time
 
+    // ── Policy server integration (PolicyControllerAi) ────────
+    // When set, the Forge command includes --policyServer so that
+    // PolicyControllerAi (in-process) or the external policy server
+    // can intercept AI decisions during batch simulations.
+    private String policyServerUrl = null;
+
     // ── Regex patterns for parsing Forge sim output ────────────────────
 
     // "Game Outcome: Turn 11"
@@ -186,6 +193,22 @@ public class BatchRunner {
         this.aiThinkTimeMs = thinkTimeMs;
     }
 
+    /**
+     * Enable policy server integration for batch simulations.
+     *
+     * When set, the Forge subprocess command includes
+     * {@code --policy midrange --policyServer <url>} flags, and
+     * the log output notes that {@link PolicyControllerAi} is
+     * available for decision interception.
+     *
+     * @param url Base URL of the Python policy server
+     *            (e.g. {@code "http://localhost:8080"})
+     */
+    public void setPolicyServer(String url) {
+        this.policyServerUrl = url;
+        System.out.println("[BatchRunner] PolicyControllerAi integration enabled — server: " + url);
+    }
+
     public void enableMlLogging(String resultsDir, String batchId) {
         this.mlLoggingEnabled = true;
         this.mlResultsDir = resultsDir;
@@ -438,6 +461,19 @@ public class BatchRunner {
         cmd.add("-c");
         cmd.add(String.valueOf(clockSeconds));
 
+        // Policy server integration (PolicyControllerAi)
+        // When policyServerUrl is set, tell Forge to use policy-driven AI.
+        // The Python policy server at this URL handles POST /api/policy/decide
+        // and returns macro-actions that PolicyControllerAi maps to concrete
+        // SpellAbility objects.
+        if (policyServerUrl != null && !policyServerUrl.isBlank()) {
+            cmd.add("--policy");
+            cmd.add("midrange");
+            cmd.add("--policyServer");
+            cmd.add(policyServerUrl);
+            System.out.println("[BatchRunner] Policy server: " + policyServerUrl);
+        }
+
         if (quiet) {
             cmd.add("-q");
         }