Find a way to run and deploy notebooks to avoid huge diffs

[matthewfeickert]

trafficstars

Summary

As noted in PR #1642, making fixes to notebooks can result in huge diffs. This makes people less inclined to want to refactor notebooks given how expensive the diffs are. Along the lines of Issue #168, it would be useful if it was possible to:

• Store empty notebooks stripped of all output
• On releases run and evaluate them in CI and then have the built versions be used in the docs build.

I'm not sure how this works, other than matplotlib has an entirely seperate build and deploy process for their docs which includes generating all the code examples.

Additional Information

No response

Code of Conduct

• [X] I agree to follow the Code of Conduct