Added LangChain_Chat module

[isaacwasserman]

This module allows you to use any existing LangChain chat model integration with Vanna.

Using this integration requires no knowledge of the LangChain APIs beyond instantiating a BaseChatModel. Additionally, the existing LLM integrations (OpenAI, Anthropic, etc.) can be replaced by minimal subclasses of LangChain_Chat; an example of this is shown below for OpenAI_Chat.

The goal of this PR is to reduce the need for maintenance of multiple LLM integrations, especially integrations with small userbases.

import os
from ..langchain import LangChain_Chat
from langchain_openai import ChatOpenAI

class OpenAI_Chat(LangChain_Chat):
    def __init__(self, config=None):
        self.temperature = config.get("temperature", 0.7)
        self.model = config.get("model", "gpt-4o-mini")
        self.api_key = config.get("api_key", os.getenv("OPENAI_API_KEY"))
        chat_model = ChatOpenAI(api_key=self.api_key, model=self.model, temperature=self.temperature)
        LangChain_Chat.__init__(self, chat_model, config=config)