Target jupyter book builds for scheduler output format

Problem

When using notebooks to generate reports, it is common to use HTML as a target output in order to share them with others. The current jupyter scheduler supports HTML outputs at a basic level (I think via nbconvert?). Many people want a more flexible and feature-rich way to combine their narrative content with their computational outputs, or they want a way to cross-link multiple notebooks that are executed as part of a report.

Proposed Solution

Jupyter Book is a build system that could generate HTML for reports that includes the power of myst markdown, the ability to insert cell outputs into notebooks text, and the ability to create much more complex outputs via sphinx extensions.

I think it could be interesting to explore how this scheduling extension could automatically build HTML via jupyter Book. This might be via executing first and then passing the executed notebooks to the scheduler, or potentially by using jupyter-cache directly from within the job as part of a jupyter book build process.

Additional context

For two examples of the kind of thing I'm talking about, here are two websites that I auto-generate via GitHub Workflows and a jupyter book build:

the 2i2c financial transparency page: https://2i2c.org/kpis/finances.html
A little personal dashboard I use to track activity across many jupyter repositories: https://chrisholdgraf.com/jupyter-activity-snapshot/executablebooks.html

Another option might be for jupyter book to have its own jupyterlab extension and API to trigger jupyter book builds via jupyterlab.

jupyter-server / jupyter-scheduler