Remove non-existent fastembed-rerank dependency

Fix memory accumulation: batch processing + periodic GC + single worker

Fix memory accumulation with batch processing and periodic GC

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Update requirements.txt

Update requirements.txt

Replace with FastEmbed ONNX models (jina-embeddings-v2-base-code + BM25 + reranker)

Rewrite README with concise model and endpoint docs

Allow extra request fields like api_key

Configure space for deployment

initial commit