Introducing Milvus SDK v2: Native Async Support, Unified APIs, and Superior Performance

Milvus SDK v2 introduces native asynchronous support (e.g., `AsyncMilvusClient` in Python) for improved concurrency in operations like batch inserts and parallel queries. A Schema Cache is implemented to store collection schemas locally, reducing network requests and CPU overhead for high-frequency operations, leading to lower latency and better scaling. The RESTful API now has near feature parity with gRPC, allowing core operations like collection management and indexing via REST. SDKs across languages are standardized around a `MilvusClient` class for a consistent developer experience. The PyMilvus `MilvusClient` simplifies collection creation, index building, and data loading into a single call. The MCP Server is introduced for AI integration, enabling AI agents to interact with Milvus clusters via the Model Context Protocol.