Considerations for new infrastructure

[dopplershift]

We've been kicking around continued rearchitecting this, so I just wanted to capture some thoughts (and close some browser tabs). The problem with the current stack, Nikola, is that it doesn't seem to buy us much while requiring a whole new set of effort to theme it up consistent with Unidata stuff. Other options:

• Jupyter Book: Works well with notebooks and markdown. Still have our theming issues. Also does not seem to have a gallery
• Sphinx with notebook extensions: Still theming, but at least a common stack to the rest of the Python projects.
• nbconvert + jekyll: Can re-use the jekyll work that's been done for e.g. netCDF-java. Downside is we'd have to continue doing all the notebook stuff on our own.