[Feature Request]: V3 - different models for different conversations in the same chatbot object

[gchust]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits

What would your feature do ?

Currently, one chatbot object can have different conversations, but all those conversations are using the same model engine.
It would be more flexible if those conversations could have a different model engine.
We can even allow user to set engine dynamic when calling the Ask API (just like the openai api).

Proposed workflow

Maybe we can create separate Conversation class, and each conversation object can manage engine/token limit by itself...

Additional information

I just found that there is a similar PR for V1: #1269

For V3, this is more complex because we should take the token limit/truncate things into consideration.