This PR significantly enhances the dataset card for `ReasonSeg_test` by adding crucial information and improving its structure.
Key changes include:
* Adding `task_categories: image-segmentation` for better discoverability.
* Including `license: cc-by-nc-4.0` and relevant `tags` such as `reasoning`, `reinforcement-learning`, `zero-shot`, `multimodal`, and `language: en`.
* Providing direct links to the associated paper ([Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcement](https://huggingface.co/papers/2503.06520)) and the official GitHub repository ([https://github.com/dvlab-research/Seg-Zero](https://github.com/dvlab-research/Seg-Zero)).
* Populating the content section with the paper abstract, an overview of the Seg-Zero framework, visual examples, and a practical sample usage guide for inference, derived from the official GitHub README.
* Adding the BibTeX citation for proper attribution.
These improvements make the dataset more informative, discov…
