⬆️ UPGRADE: sphinx v4 (& drop sphinx v2, change mathjax override)

[chrisjsewell]

The main change is with mathjax. It's still a bit of a faff, but I think the approach is a little better:
tex2jax_process and mathjax_process classes are added to the top-level sections of myst documents (this "fixes" having to override setting for RST documents), this tells mathjax not to render for any math under this section (i.e. the entire document) by default. The math class is then added to the process list, which overrides the default, and so tells mathjax to render elements (and child elements) with this class.
So basically it constricts mathjax to only rendering the elements we want.
This is also explained at; https://myst-parser--390.org.readthedocs.build/en/390/using/syntax.html#mathjax-and-math-parsing

As well as this, most of the sphinx test regression files have been split into sphinx3 (docutils<0.17) and sphinx4 (docutil 0.17) variants.

closes #378

[codecov]

Codecov Report

Merging #390 (df1edfa) into master (5f37c0b) will decrease coverage by 0.15%.
The diff coverage is 75.75%.

@@            Coverage Diff             @@
##           master     #390      +/-   ##
==========================================
- Coverage   90.54%   90.39%   -0.16%     
==========================================
  Files          14       14              
  Lines        1830     1843      +13     
==========================================
+ Hits         1657     1666       +9     
- Misses        173      177       +4     

Flag	Coverage Δ
pytests	90.39% <75.75%> (-0.16%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
myst_parser/mathjax.py	80.48% <50.00%> (-16.39%)	⬇️
myst_parser/main.py	87.50% <80.00%> (-0.44%)	⬇️
myst_parser/docutils_renderer.py	93.17% <100.00%> (+0.65%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5f37c0b...df1edfa. Read the comment docs.

[mmcky]

🥳 -- thanks @chrisjsewell. Will do some testing. What's your timeframe for feedback (i.e. when do you want to merge this?)

[chrisjsewell]

Well ASAP I guess for those waiting on it, can you look in the next day or 2?
Also @choldgraf?

[choldgraf]

This looks great to me, just one minor under-the-hood thought but it's no biggie. Exciting to see support for v4!

Also I believe this will close #378 so added that to the top comment

[chrisjsewell]

Also I believe this will close #378 so added that to the top comment

FYI this was already in the linked issues. In case you didn't know adding in the initial comment is only useful when you first create the PR; to auto-add to the linked issues, after that it doesn't do anything