deploy Allergy-Safe Recipe Assistant

fix: try Fireworks AI for vision routing

fix: route vision through Together AI (supports Llama 3.2 Vision)

fix: route vision calls through Groq provider (free Llama 3.2 Vision)

fix: use Llama-3.2-11B-Vision via Together AI for image extraction

fix: switch chat model to Qwen2.5-7B-Instruct (Mistral-v0.3 not chat-compatible on serverless API)

fix: switch vision model to Qwen2-VL (Llama requires provider enablement)

deploy Allergy-Aware Recipe Classifier to HF Spaces

Duplicate from streamlit/streamlit-template-space