Openai-compatible embeddings

[Wolframko]

Description

Adding support for OpenAI-compatible embedding servers would be a valuable enhancement to the project. This would enable integration with local embedding models and services that implement the OpenAI embeddings API format.
I'm willing to implement this feature myself if it aligns with the project's direction and roadmap.

Acceptance Criteria

• Support for configurable OpenAI-compatible embedding endpoint (base URL)
• API key authentication support for embedding requests
• Model selection parameter for embedding generation
• Batch embedding support (multiple texts in single request)
• Error handling for embedding service failures
• Documentation with configuration examples
• Backward compatibility with existing embedding implementations

Additional Context

This feature would complement the existing RAG infrastructure by providing flexibility in embedding model selection without vendor lock-in.

[siwardean]

I would even extend it to support OpenAI compatible providers with configurable base url endpoint