Add model card: upstream credits, roles, licenses

Mirror the core-resources embedding ONNX for reproducible builds

Mirror paraphrase-multilingual-MiniLM-L12-v2-tokenizer.json so a clone can build

Mirror minilm-tokenizer.json so a clone can build

Mirror minilm-l6-v2-q8.onnx so a clone can build

Add SetFit routing encoder (q8 ONNX) so the build can fetch it

Mirror paraphrase-l12 (bundled default embedding model)

Mirror paraphrase-l12 (bundled default embedding model)

initial commit