feat(adversarial): flat_too_long_threshold parametrico (CLI ablation)

Estende AdversarialAgent con flat_too_long_threshold (default 0.95)
configurabile, simmetrico a fees_eat_alpha_threshold. Propagato a
RunConfig.flat_too_long_threshold e flag CLI --flat-too-long-threshold.

Motivazione: pop30-combo ha registrato 75 finding flat_too_long HIGH
(secondo killer dopo fees_eat_alpha 87). Rilassare la soglia 0.95→0.98
ammette strategie più passive ma marginalmente attive — analogo
all'ablation fees già verificata (+23% stabile).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/multi_swarm/agents/adversarial.py

@@ -63,9 +63,11 @@ class AdversarialAgent:
         self,
         fees_bp: float = 5.0,
         fees_eat_alpha_threshold: float = 0.5,
+        flat_too_long_threshold: float = 0.95,
     ) -> None:
         self._engine = BacktestEngine(fees_bp=fees_bp)
         self._fees_eat_alpha_threshold = fees_eat_alpha_threshold
+        self._flat_too_long_threshold = flat_too_long_threshold
 
     def review(self, strategy: Strategy, ohlcv: pd.DataFrame) -> AdversarialReport:
         signal_fn = compile_strategy(strategy)
@@ -133,12 +135,15 @@ class AdversarialAgent:
         n_active = int(((signals == Side.LONG) | (signals == Side.SHORT)).sum())
         n_flat_or_nan = n_bars - n_active
         flat_ratio = n_flat_or_nan / n_bars if n_bars > 0 else 1.0
-        if flat_ratio > 0.95:
+        if flat_ratio > self._flat_too_long_threshold:
             report.findings.append(
                 Finding(
                     name="flat_too_long",
                     severity=Severity.HIGH,
-                    detail=f"Signal flat for {flat_ratio * 100:.1f}% of bars (>95% threshold)",
+                    detail=(
+                        f"Signal flat for {flat_ratio * 100:.1f}% of bars "
+                        f"(>{self._flat_too_long_threshold * 100:.0f}% threshold)"
+                    ),
                 )
             )