feat: extend context window to 3 months (2,160 hours) for 96GB VRAM

- Updated context_hours from 1,440 to 2,160 across all modules
- Version bump to v1.6.0
- Optimized for upgraded hardware: 96GB VRAM, 48 vCPU, 186GB RAM
- Previous OOM issues resolved with hardware upgrade
- Memory profiling diagnostics remain active for baseline analysis

Hardware: 48 vCPU, 186 GB RAM, 96 GB VRAM
Context: 2,160 hours (90 days / 3 months)
Files: app.py, chronos_inference.py, dynamic_forecast.py

app.py

@@ -2,8 +2,8 @@
 """
 FBMC Chronos-2 Forecasting API
 HuggingFace Space Gradio Interface
-Version: 1.5.3 - Extended Context Window (1,440 hours = 60 days)
-FORCE REBUILD: GPU memory optimized context for 48GB hardware
+Version: 1.6.0 - Extended Context Window (2,160 hours = 90 days / 3 months)
+FORCE REBUILD: Optimized for 96GB VRAM with memory profiling diagnostics
 """
 
 # CRITICAL: Set PyTorch memory allocator config BEFORE any imports

src/forecasting/chronos_inference.py

@@ -3,7 +3,7 @@
 Chronos-2 Inference Pipeline with Past-Only Covariate Masking
 Standalone inference script for HuggingFace Space deployment.
 Uses predict_df() API with ALL 2,514 features leveraging Chronos-2's mask-based attention.
-FORCE REBUILD: v1.5.3 - Extended context window (1,440 hours = 60 days) optimized for 48GB GPU
+FORCE REBUILD: v1.6.0 - Extended context window (2,160 hours = 90 days) optimized for 96GB VRAM
 """
 
 import os
@@ -128,7 +128,7 @@ class ChronosInferencePipeline:
         run_date: str,
         borders: Optional[List[str]] = None,
         forecast_days: int = 7,
-        context_hours: int = 1440,  # 1,440 hours = 60 days (GPU memory optimized for 48GB)
+        context_hours: int = 2160,  # 2,160 hours = 90 days (3 months, optimized for 96GB VRAM)
         num_samples: int = 20
     ) -> Dict:
         """

src/forecasting/dynamic_forecast.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 """
-Dynamic Forecast Module v1.5.3 - Extended Context Window (60 Days)
+Dynamic Forecast Module v1.6.0 - Extended Context Window (90 Days / 3 Months)
 Time-aware data extraction for forecasting with run-date awareness.
 
 Purpose: Prevent data leakage by extracting data AS IT WAS KNOWN at run time.
@@ -8,7 +8,7 @@ Purpose: Prevent data leakage by extracting data AS IT WAS KNOWN at run time.
 Key Concepts:
 - run_date: When the forecast is made (e.g., "2025-09-30 23:00")
 - forecast_horizon: Always 14 days (D+1 to D+14, fixed at 336 hours)
-- context_window: Historical data before run_date (1,440 hours = 60 days)
+- context_window: Historical data before run_date (2,160 hours = 90 days / 3 months)
 - future_covariates: ALL 2,514 features (leveraging Chronos-2 past-only masking)
   * 603 full-horizon (known future)
   * 12 partial D+1 (masked D+2-D+14)
@@ -39,7 +39,7 @@ class DynamicForecast:
     def __init__(
         self,
         dataset: pl.DataFrame,
-        context_hours: int = 1440,  # 1,440 hours = 60 days (GPU memory optimized for 48GB)
+        context_hours: int = 2160,  # 2,160 hours = 90 days (3 months, optimized for 96GB VRAM)
         forecast_hours: int = 336  # Fixed at 14 days
     ):
         """