Add support for LLama CPP

[vmesel]

In order for us to enable users to run their finetuned models that are LlamaCPP compatible, we need to add this compatibility in our class structure.

[lgabs]

Nice, local models can be very useful for local tasks like prompt evaluations or even for the main task of Q&A. Langchain seams to have the integration with llama.cpp.