Better `@generative` documentation in the tutorial

[nrfulton]

The tutorial "jumps right in" too quickly.

We should start with a simple one-function example. We should demonstrate how you can swap out the model and how to specify model_options. Sketch below.

How it Works

@generative
def my_function(...) -> ...:
    """ .... """

results in a GenerativeSlot object, which has the following call signature (all the uppercase letters are generic types):

    def __call__(
        self, m, model_options: dict | None = None, *args: P.args, **kwargs: P.kwargs
    ) -> R:

[[TODO: point to code repo]]

where m: MelleaSession. Sessions hold a Backend which has a model_id. You can pass whatever session you want into a generative slot. So, via the first parameter to any generative slot, you are able to specify the LLM.
The model_options dict allows you to additionally control params.
We should start more gadually in the tutorial, pointing out these options.

[psschwei]

we've updated tutorial