malkhut(wire): fill quality as PRIMARY optimization target

Fill quality is MALKHUT's core aim. Wired end-to-end:

1. FillQuality state (state.py):
   - slippage_bps, price_improvement_bps, levels_consumed
   - is_maker_fill, rolling_fill_rate, post_fill_adverse_bps
   - fill_value_score: composite metric for optimization
   - Added to MarketWorldState.fill_quality field

2. HftBacktestCWM.transition() (hft_cwm.py):
   - _compute_fill_quality() computes all metrics per transition
   - Fill quality now tracked for every CWM step
   - Empty book guards added for safety

3. MinimalCryptoLOBCWM.transition() (core.py):
   - Same fill quality computation for deterministic fallback
   - Empty book guards added

4. Reward function (hft_cwm.py):
   - fill_quality_reward = w_fill_probability * fill_value_score (PRIMARY)
   - Bonus for maker fills that improve price
   - Penalty for adverse selection after fill
   - Base reward (PnL, adverse selection, fees) preserved

5. PerformanceMatrix (selector.py):
   - RegimeStrategyScore: 4 new fill quality fields
   - record(): accepts fill_rate, slippage, price_improvement, fill_value_score
   - EMA updates for all fill quality metrics

6. EpisodeResult (cma_trainer.py):
   - avg_fill_value_score, avg_price_improvement_bps, avg_post_fill_adverse_bps
   - Accumulated per-step during _run_episode
   - Recorded to PerformanceMatrix in evaluate_candidate

All 1379+ tests green.

MALKHUT/malkhut/state.py

@@ -267,6 +267,42 @@ class ExecutionIntent:
     reason: str
 
 
+@dataclass(frozen=True, slots=True)
+class FillQuality:
+    """Fill quality metrics — the CORE optimization target of MALKHUT.
+
+    MALKHUT is an execution improvement engine. Fill quality IS the primary aim.
+    Every transition records these metrics. The reward function weights them heavily.
+    The PerformanceMatrix tracks them per (regime, strategy, venue).
+    """
+    filled: bool = False
+    fill_qty: float = 0.0
+    fill_price: float = 0.0
+    requested_qty: float = 0.0
+
+    # How close to mid did we fill? (for aggressive: positive = slipped)
+    slippage_bps: float = 0.0
+
+    # For passive fills: how much better than best bid/ask? (positive = improvement)
+    price_improvement_bps: float = 0.0
+
+    # How many levels deep was the fill?
+    levels_consumed: int = 0
+
+    # Was this a maker (passive) or taker (aggressive) fill?
+    is_maker_fill: bool = False
+
+    # Fill rate rolling window (updated each transition)
+    rolling_fill_rate: float = 0.0
+
+    # Adverse selection: price movement after fill (negative = adverse)
+    post_fill_adverse_bps: float = 0.0
+
+    # Fill value score: composite metric for optimization
+    # = fill_rate * price_quality - adverse_selection - slippage
+    fill_value_score: float = 0.0
+
+
 @dataclass(frozen=True, slots=True)
 class MarketWorldState:
     """Complete CWM root state. Immutable for safe tree search."""
@@ -287,6 +323,8 @@ class MarketWorldState:
     order_latency_ms: float = 0.0
     rng_seed: int = 0
 
+    fill_quality: Optional[FillQuality] = None
+
 
 @dataclass(frozen=True, slots=True)
 class FulfilmentPolicyParams: