fix(ml): complete ML pipeline fixes and setup

- Fix CCI indicator to use HLC prices instead of close only
- Parse datetime column when loading enriched CSV
- Strip timezone from annotation timestamps
- Fix TA-Lib pattern names (CDL3WHITESOLDIERS, CDL3BLACKCROWS)
- Exclude programmatic label columns from training features
- Fix classification report to handle missing classes
- Update MLflow tracking to use localhost:5000
- Grant PostgreSQL permissions to ml_user

Pipeline now runs successfully end-to-end:
- Feature engineering: 2543 rows, 31 columns
- Annotation ingestion: 286 samples
- Training: 89.47% test accuracy with Random Forest

services/ml/mlruns/358560345319124639/f1a0c7b223484d019564929940766c10/artifacts/classification_report.txt

@@ -0,0 +1,8 @@
+                   precision    recall  f1-score   support
+
+Bearish Engulfing     0.8065    1.0000    0.8929        25
+Bullish Engulfing     1.0000    0.8125    0.8966        32
+
+         accuracy                         0.8947        57
+        macro avg     0.9032    0.9062    0.8947        57
+     weighted avg     0.9151    0.8947    0.8949        57

services/ml/mlruns/358560345319124639/f1a0c7b223484d019564929940766c10/artifacts/pipeline_config.yaml

@@ -0,0 +1,97 @@
+data:
+  annotations_path: data/annotations/export.json
+  enriched_path: data/enriched/features.csv
+  labeled_path: data/labeled/dataset.csv
+  raw_path: data/raw/OHLCV.csv
+stages:
+  annotation_ingestion:
+    context_padding: 20
+    enabled: true
+    label_encoding: window
+    merge_strategy: human_priority
+    min_confidence: 1
+    programmatic_labels:
+      enabled: true
+      talib_patterns:
+      - CDLENGULFING
+      - CDLHAMMER
+      - CDLINVERTEDHAMMER
+      - CDLSHOOTINGSTAR
+      - CDLDOJI
+      - CDLDOJISTAR
+      - CDLMORNINGSTAR
+      - CDLEVENINGSTAR
+      - CDLHARAMI
+      - CDLPIERCING
+      - CDLDARKCLOUDCOVER
+      - CDL3WHITESOLDIERS
+      - CDL3BLACKCROWS
+    window_size: 30
+  feature_engineering:
+    candle_features: true
+    custom_features: []
+    enabled: true
+    talib_indicators:
+    - name: RSI
+      params:
+        timeperiod: 14
+    - name: EMA
+      params:
+        timeperiod: 20
+    - name: EMA
+      params:
+        timeperiod: 50
+    - name: MACD
+      params:
+        fastperiod: 12
+        signalperiod: 9
+        slowperiod: 26
+    - name: BBANDS
+      params:
+        nbdevdn: 2
+        nbdevup: 2
+        timeperiod: 20
+    - name: ATR
+      params:
+        timeperiod: 14
+    - name: ADX
+      params:
+        timeperiod: 14
+    - name: CCI
+      params:
+        timeperiod: 14
+    - name: MFI
+      params:
+        timeperiod: 14
+    - name: STOCH
+      params:
+        fastk_period: 14
+        slowd_period: 3
+        slowk_period: 3
+  inference:
+    batch_size: 1000
+    enabled: true
+    local_model_path: models/best_model.pkl
+    mlflow_model_name: candlestick_pattern_v1
+    mlflow_model_stage: Production
+    model_source: local
+    use_training_config: true
+  training:
+    class_weights: balanced
+    enabled: true
+    hyperparameters:
+      max_depth: 15
+      min_samples_leaf: 2
+      min_samples_split: 5
+      n_estimators: 200
+      n_jobs: -1
+      random_state: 42
+    mlflow:
+      experiment_name: candlestick_patterns
+      log_artifacts: true
+      register_model: false
+      tracking_uri: http://localhost:5000
+    model_type: random_forest
+    split_method: temporal
+    test_split: 0.2
+    validation_split: 0.1