(REFAC)[Enhance ResultsManager]: Expand ResultsManager for multi-model session tracking and export

- Refactored `ResultsManager` to support robust session-wide results management for multi-file and multi-model inference.
- Added methods for initializing, adding, retrieving, and clearing results in Streamlit session state.
- Implemented spectrum data retrieval for files, DataFrame conversion, and export to CSV/JSON.
- Enhanced summary statistics and accuracy calculation, including ground truth support.
- Added multi-model results aggregation and comparative statistics functions.
- Implemented agreement matrix calculation for model prediction consistency.
- Created comprehensive comparison visualization and exportable JSON report.
- Improved ephemeral state reset for full app refresh.
- Enhanced results table display with metrics, export buttons, and dashboard page linking.
- Improved error handling, modularity, and extensibility for future

utils/results_manager.py

@@ -12,6 +12,7 @@ from pathlib import Path
 import io
 from collections import defaultdict
 import matplotlib.pyplot as plt
+from matplotlib.figure import Figure
 
 
 def local_css(file_name):
@@ -321,8 +322,7 @@ class ResultsManager:
 
         return pd.DataFrame(agreement_matrix, index=all_models, columns=all_models)
 
-    @staticmethod
-    def create_comparison_visualization() -> plt.Figure:
+    def create_comparison_visualization() -> Figure:
         """Create visualization comparing model performance."""
         comparison_stats = ResultsManager.get_comparison_stats()