chore: remove deprecated FeatureVector54 type alias

Dimension was reduced from 54 to 51 in WAVE 10. All usages now
use FeatureVector ([f64; 51]) directly.

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

common/src/features/types.rs

@@ -3,10 +3,6 @@
 /// Standard 51-dimensional feature vector (Proxy OFI removed in WAVE 10)
 pub type FeatureVector51 = [f64; 51];
 
-/// DEPRECATED: 54-dimensional feature vector (backward compatibility only)
-#[deprecated(since = "WAVE 10", note = "Use FeatureVector51 instead")]
-pub type FeatureVector54 = [f64; 54];
-
 /// OHLCV bar data for batch processing
 #[derive(Debug, Clone)]
 pub struct BarData {

common/src/lib.rs

@@ -82,9 +82,9 @@ pub use ml_strategy::{
 // Re-export regime persistence manager
 pub use regime_persistence::RegimePersistenceManager;
 
-// Re-export feature extraction types and functions (54-feature architecture)
+// Re-export feature extraction types and functions
 pub use features::{
-    FeatureVector54, BarData,
+    BarData,
     // Technical indicators (streaming + batch)
     RSI, EMA, MACD, BollingerBands, ATR, ADX,
     rsi_batch, ema_batch, macd_batch, bollinger_batch, atr_batch, adx_batch,

ml/src/features/extraction.rs

@@ -60,10 +60,6 @@ pub struct OHLCVBar {
 /// Feature extraction result: 51-dimensional feature vector per bar
 pub type FeatureVector = [f64; 51];
 
-/// DEPRECATED: 54-dimensional feature vector (backward compatibility only)
-#[deprecated(since = "WAVE 10", note = "Use FeatureVector (51 dimensions) instead")]
-pub type FeatureVector54 = [f64; 54];
-
 /// Base feature vector without OFI: 43-dimensional feature vector
 pub type FeatureVector43 = [f64; 43];
 

ml/src/lib.rs

@@ -1805,7 +1805,6 @@ pub struct LatencyOptimizer {
 }
 
 #[derive(Debug, Clone)]
-#[allow(dead_code)]
 struct PerformancePoint {
     timestamp: std::time::Instant,
     latency_us: u64,
@@ -1815,7 +1814,6 @@ struct PerformancePoint {
 }
 
 #[derive(Debug, Clone)]
-#[allow(dead_code)]
 struct OptimizationParams {
     max_batch_size: u32,
     adaptive_batching: bool,

ml/src/mamba/mod.rs

@@ -1625,7 +1625,6 @@ impl Mamba2SSM {
     }
 
     /// Forward pass with gradient computation enabled
-    #[allow(dead_code)] // Used in tests
     pub fn forward_with_gradients(&mut self, input: &Tensor) -> Result<Tensor, MLError> {
         // Gradient flow enabled - do not detach
         let input = input;
@@ -1882,7 +1881,6 @@ impl Mamba2SSM {
     }
 
     /// Compute training loss
-    #[allow(dead_code)] // Used in tests
     pub fn compute_loss(&self, output: &Tensor, target: &Tensor) -> Result<Tensor, MLError> {
         // Mean Squared Error for regression
         let diff = (output - target)?;
@@ -1893,7 +1891,6 @@ impl Mamba2SSM {
     }
 
     /// Backward pass - compute gradients for model parameters
-    #[allow(dead_code)] // Used in tests
     pub fn backward_pass(
         &mut self,
         loss: &Tensor,
@@ -1992,7 +1989,6 @@ impl Mamba2SSM {
     }
 
     /// Zero gradients
-    #[allow(dead_code)] // Used in tests
     pub fn zero_gradients(&mut self) -> Result<(), MLError> {
         // Clear all gradients for SSM parameters
         for _ssm_state in &mut self.state.ssm_states {

ml/src/ppo/unified_ppo.rs

@@ -5,13 +5,10 @@
 
 // Placeholder types to satisfy mod.rs exports
 #[derive(Debug)]
-#[allow(dead_code)]
 pub struct UnifiedPPO;
 
 #[derive(Debug)]
-#[allow(dead_code)]
 pub struct UnifiedPPOConfig;
 
 #[derive(Debug)]
-#[allow(dead_code)]
 pub struct UnifiedTrajectoryBatch;

ml/tests/dqn_call_sites_remaining_test.rs

@@ -6,8 +6,6 @@
 use ml::dqn::portfolio_tracker::PortfolioTracker;
 use ml::dqn::trading_action::TradingAction;
 use ml::trainers::dqn::{DQNHyperparameters, DQNTrainer};
-use ml::types::FeatureVector54;
-
 #[tokio::test]
 async fn test_process_training_sample_call_sites() {
     // Test that process_training_sample() correctly passes close_price to feature_vector_to_state()