Add logging to silent ML energy prediction fallback

src/data/data_providers.py

@@ -964,8 +964,8 @@ class EnergyService(BaseService):
                 from src.energy_predictor import EnergyPredictor
                 ep = EnergyPredictor()
                 return ep.predict_day_from_weather_df(target_date, df.tail(96))
-            except Exception:
-                pass  # fall through to raw analytical
+            except Exception as exc:
+                log.warning("ML energy prediction unavailable, falling back to analytical: %s", exc)
 
             # Raw analytical: GHI × capacity / STC
             day_df = df.tail(96).copy()