This PR significantly enhances the dataset card for MultiTableQA by:
- Adding `task_categories: ['table-question-answering']` to the metadata for improved discoverability.
- Incorporating a detailed introduction about the T-RAG framework and MultiTableQA benchmark, directly sourced from the associated GitHub repository README.
- Adding a comprehensive "Sample Usage" section, including code snippets for installation, data preparation, T-RAG retrieval stages, and downstream inference with LLMs, all extracted from the official GitHub README.
- Correcting the `title` field in the BibTeX citation to accurately reflect the paper's title: "RAG over Tables: Hierarchical Memory Index, Multi-Stage Retrieval, and Benchmarking".
These updates provide a much clearer and more actionable overview of the dataset and its usage for the Hugging Face community.
