Feature request

I'd be interested to implement a Python backend for the classifier. I would be up to work on this.

Motivation

The Python backend allows for prototyping models with trust_remove_code=True faster. Would be great to see the router gauge possible performance impact when performing heavy python operations thanks to low grcp overhead.

Your contribution

I would like to add / modify the following files:

backend/Proto: Add Predict to EmbeddingService in https://github.com/huggingface/text-embeddings-inference/tree/main/backends/proto
backend/grcp-client: Add a pub asyn predict() in https://github.com/huggingface/text-embeddings-inference/tree/main/backends/grpc-client/src
backend/python/libs.rs: Add the logic for fn predict() in https://github.com/huggingface/text-embeddings-inference/blob/main/backends/python/src/lib.rs
Python: https://github.com/huggingface/text-embeddings-inference/blob/main/backends/python/server/text_embeddings_server/server.py

please let me know if you have a preferred choice.