Add guide on executable documentation

[danielballan]

Following up #136, investigate whether we should recommend:

• Jupytext
• the ipython sphinx directive
• the matplotlib plot directive

I suspect that Jupytext is now the best way to do this, and more compatible with a Markdown-based approach, but I want to educate myself more.

[danielballan]

I started looking at this more today, aligned with a day-job task. I intend to put up a PR on this in the next week or so.

[danielballan]

Following up here to mention, belatedly, that I have a self-contained demo repo illustrating some nice patterns for interactive documentation:

• source in Markdown (not ipynb JSON)
• auto-generated branch with ipynb JSON
• published HTML with executed code
• Binder
• Jupyter Lite (proof of concept)

https://github.com/danielballan/interactive-tutorial-demo

I could imagine promoting this to live in the scientific-python org if there is support and interest.

[danielballan]

Given the range of technologies and use cases, I think it would be too heavy to include in this repository, so I would advocate for linking out. Most major projects have a dedicated "examples" repo separate from the main library to do things like this.