You are selecting a vector database for a production RAG system with 100M vectors, 10,000 QPS, multi-tenant requirements, and a team of 3 engineers. Compare the major options: Pinecone, Weaviate, Qdrant, Chroma, pgvector, OpenSearch, and Milvus. For each, explain the key strengths, operational overhead, query performance characteristics, cost model, and when it is the wrong choice. What is your decision framework? Which would you choose for this specific scenario and why?