Vector, full-text, and hybrid search on a fully disaggregated architecture. Compute, memory, and storage scale independently with pools: serverless, fast at any size, 10x lower cost