docs: fix Limitations — audio duration instead of prompt length

docs: add Tested inference path section, reorganize model card

card: full rewrite from canonical template

card: append canonical Vibecrafted footer

card: add 'Inference tested on' link to mlx-batch-server

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

initial commit