Adding more in depth docs/tutorials

[nihargupte-ph]

The goal of this PR is to add 3 tutorials to docs/source.

Tutorial 1: Toy Example

This should be able to run on a local machine in < 10 minutes. The goal here is to just explain some settings about the code and what it is outputting

Tutorial 2: Normalizing Flow without GNPE

This would contain production settings to train a normalizing flow. The goal here is to explain the production settings and how to change them to suit the user's needs. The problem with this tutorial is I have not trained a network without GNPE for a very long time, so actually, I'm not sure what the optimal settings are. For now, I was just going to put the same settings as GNPE.

Tutorial 3: Normalizing Flow with GNPE

Same as tutorial 2, except including the GNPE. These would be the recommended settings.

Note it is still a draft at the moment since need to fix some grammar and add more tutorials but feel free to add things.

[stephengreen]

Thanks Nihar. Regarding naming, I think it would be good to call the examples something like

• Toy NPE model
• NPE model (production)
• GNPE model (production)

That way we avoid naming things by what they are not, e.g., "No GNPE".

[mpuerrer]

I have make some edits to the toy example tutorial, mostly minor things to make the wording more concise. Let me know if there is anything you disagree with. Will have a look at the other tutorials next.

[stephengreen]

It also needs to be decided how to integrate these examples together with the "quickstart tutorial", which has significant overlap. Maybe, e.g., the quickstart could be replaced by the toy example, but the flow chart could go in the GNPE example. A simplified flow chart could also be used for the two NPE examples.

This should also be addressed at some point, but I understand if there isn't enough time now.