Fix MLflow localhost references to use Docker service name

Replace hardcoded localhost:5000 with mlflow:5000 in pipeline.yaml
and main.py health check fallback so containers can reach MLflow
over Docker's internal network.

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

services/ml/app/main.py

@@ -536,7 +536,7 @@ async def health_check(response: Response):
     # Check MLflow connection via HTTP GET to its health endpoint
     mlflow_status = "unhealthy"
     try:
-        mlflow_tracking_uri = os.getenv("MLFLOW_TRACKING_URI", "http://localhost:5000")
+        mlflow_tracking_uri = os.getenv("MLFLOW_TRACKING_URI", "http://mlflow:5000")
         mlflow_health_url = f"{mlflow_tracking_uri.rstrip('/')}/health"
         resp = http_requests.get(mlflow_health_url, timeout=3)
         if resp.status_code == 200:

services/ml/config/pipeline.yaml

@@ -127,7 +127,7 @@ stages:
     
     # MLflow settings
     mlflow:
-      tracking_uri: "http://localhost:5000"
+      tracking_uri: "http://mlflow:5000"
       experiment_name: "candlestick_patterns"
       log_artifacts: true
       register_model: false  # Set to true to register in model registry