malkhut: configurable friction per scenario + 5h learning test

MALKHUT/malkhut/long_learning_5h.py

@@ -52,7 +52,7 @@ ASSETS = ["BTCUSDT", "ETHUSDT", "SOLUSDT", "DOGEUSDT", "ADAUSDT",
           "AVAXUSDT", "UNIUSDT", "LINKUSDT", "BNBUSDT"]
 STEPS_PER_EPISODE = 20
 SEED = 42
-N_OPPONENTS = 1500
+N_OPPONENTS = 500
 
 
 class ResourceMonitor:
@@ -248,6 +248,11 @@ def main():
         except Exception as e:
             print(f"  Error: {e}", flush=True)
 
+        # Memory management: gc.collect every 100 episodes
+        if fq_n_reports % 100 == 0:
+            import gc
+            gc.collect()
+
         # CMA-ES every 15 reports
         if fq_n_reports % 15 == 0 and remaining > 600:
             try: