simple API interface for initializing and running model inference

[aniketmaurya]

We want to serve LLMs from LitGPT using LitServe, however the current model initialization step leaks a lot of complexity at the user code. Also, couldn't find a generator function to stream responses. So had to bring the generate function on user code side.

A simple API to do this kinds of thing in a few lines of code would be really appreciated!

cc: @lantiga