This PR adds a comprehensive model card for NativeTok. It includes:
- Metadata with the `image-to-image` pipeline tag.
- A link to the research paper: [NativeTok: Native Visual Tokenization for Improved Image Generation](https://huggingface.co/papers/2601.22837).
- A link to the official GitHub repository.
- A table of available model checkpoints.
- Sample usage for downloading the model weights using the `huggingface_hub` library.
