Support for custom OpenAI base path

[bioshazard]

I am hosting a llama-cpp-python[server] instance on my localhost. I request that support be added to specify a custom base path like http://localhost:8080, eg:

I tried to get this working myself, but initial attempt seemed to be complicated by multiple places where the OpenAI instance is instantiated. Thanks!

[bioshazard]

I was actually able to get the chat working with my local service after adding a baseURL override in the configuration of the ChatOpenAI instantiation in chat-model.ts, will push later.

[bioshazard]

btw @mabry1985 idk about you but I feel like this is delivered. I have moved to Acai as my daily chat driver with MistralOrca

[mabry1985]

@bioshazard yup, thanks! Nice job here 🥇