Another take at a Tags domain

[melissawm]

Here's an attempt at a domain approach - building on the idea by @story645. This is very similar to #92 but I am explicitly trying to do the bare minimum.

This works well to read and generate tags. The problem is:

• To create the pages for each tag and the tags index, all files must be read first
• Sphinx complains about not finding the pending references (for example, we link to _tagindex from a toctree, but this file hasn't been created yet before all files are read and tags are parsed)
• Once files are read, I can't find a way to add the new files to the source tree so they are included in the build.

Any tips are appreciated 🙏🏻

[melissawm]

Also: this actually works without the tags:tags problem so we can also try to join the best of both PRs if needed? 😄 I didn't want to do the Index part unless absolutely necessary because it sounds like it will include the tags in the global sphinx index and I'm not sure that gives us the flexibility to customize the index page? Maybe I'm wrong, I haven't tried very hard. Will come back to this in the next few days.

[melissawm]

Here's a pointer someone shared at the writethedocs slack: https://github.com/dernorberto/sphinxPageProperties/blob/main/extension/page_properties_report.py and also https://github.com/sphinx-doc/sphinx/issues/11562

[melissawm]

I think this is almost there. A couple of details left, as mentioned in the commit message:

This generates a couple of build warnings, because some references are missing on the first pass. This is expected, but we should either silence these errors or defer resolution of these references until later.

This also doesn't include badges for similar reasons: because the references in the badges can't be resolved in the first pass, the badges are not linked to any pages.

[melissawm]

Not sure why the docs build is failing on readthedocs as I don't see this failure locally. Will investigate later.