Using native upsert method for Milvus vector store

[Mateusz-Switala]

Based on the source code, current langchain_milvus.Milvus.upsert methods delete the chunks with the proper ids from the collection and then insert the chunks provided in the method: https://github.com/langchain-ai/langchain-milvus/blob/main/libs/milvus/langchain_milvus/vectorstores/milvus.py#L1993-#L2024

However, since the last few Milvus versions, there is a native upsert method for MilvusClient: https://milvus.io/api-reference/pymilvus/v2.5.x/MilvusClient/Vector/upsert.md Using the dedicated method for the upsert process, may be a more efficient approach and will definitely reduce the code complexity.

Shouldn't the langchain_milvus.Milvus use the mative upsert method under the hood?

[zc277584121]

@Mateusz-Switala thanks, you are right, we will update it within few days

[zc277584121]

fixed