Upgrade dataset analysis and visualization with extended metadata and interactive UI

[Irozuku]

This pull request introduces a major upgrade to the dataset analysis and visualization features in DashAI. The backend now computes and stores extended metadata for datasets, including detailed statistical summaries and quality indicators. The frontend has been refactored to present this information in a more interactive, tabbed interface, allowing users to explore different aspects of their datasets such as overview, numeric and categorical analysis, data quality, and correlations. Additionally, the frontend adds new dependencies and UI improvements for enhanced visualization.

Backend: Extended Metadata Computation and Storage

• The method nan_per_column is replaced by compute_metadata, which now calculates and stores comprehensive metadata for each dataset, including NaN counts, column types, numeric/categorical/text stats, quality indicators, and correlations, all in self.splits for frontend visualization.
• Metadata saving and retrieval logic is updated to ensure the new extended metadata is correctly persisted and loaded for frontend use.
• The job pipeline now calls compute_metadata instead of the old NaN calculation method, ensuring all new metadata is available after dataset processing.

Frontend: Visualization Refactor and UI Enhancements

• The dataset visualization component is refactored to use a tabbed interface (Tabs, Tab), with dedicated tabs for Overview, Numerical Analysis, Categorical, Data Quality, and Correlations, each powered by the new backend metadata.
• The quick stats and quality score alert are improved, providing users with a clear summary of dataset health and key indicators at a glance.
• New UI components and dependencies are added (e.g., recharts for charts, new MUI icons and controls) to support richer data presentation and interactivity.

Video_251111121229.mp4