GPU-Bridge + LlamaIndex: Embeddings and Reranking in One Line

Drop-in LlamaIndex integration for GPU-Bridge. High-throughput embeddings at $0.00002/call and semantic reranking for RAG pipelines.