Add ChatGPT fine-tuned model option

[ldhoax]

[mkorpela]

@ldhoax thanks for the PR.
I think the main point would be to be able to configure the model (specifically model name) used?

Current approach in OpenGPTs is a bit rigid. I hope that will be changed in the future to a configuration that live in the database.

[ldhoax]

I agree that the current approach is a bit rigid 😄 but I think it's a bit inconvenient and difficult to work with just config in the database, for several reasons:

• With new agents (like google_agent, xml_agent), we still need new code to handle
• And in my case, I just used one OpenAI key for my fine-tuned model, I think it doesn't really need a database for that key 😄 and also a bit insecure.

I think our expectation is just to add new model name to an existing object (like AgentType) every time a new model becomes available. How do you think @mkorpela 🤔