Use batch requests for embeddings for Ollama

Currently Ollama loops and manually calls embedding endpoint one at a time.

Ollama has added native support for batch embeddings. https://github.com/ollama/ollama/pull/5127

Update langchain-rust to directly use batch embeddings api when ollama-rs supports it. https://github.com/pepperoni21/ollama-rs/issues/58

Find the Ollama embedding calls in the codebase, likely in a module handling Ollama integration. Wait for ollama-rs to support batch embeddings (track issue #58). Then update the code to replace sequential calls with a batch request, ensuring the API change is compatible. Test with the existing embedding tests to verify functionality and performance improvement.

由索引模型根据 Issue 内容生成。