Fix malformed $\LaTeX$

[NathanielBlairStahn]

Fix a bunch of errors in our displayed equations, with the help of GitHub Copilot. The two most common errors were

• Unescaped underscores within a \text command, e.g., \text{deaths_c500} instead of \text{deaths\_c500}, leading to a "'_' allowed only in math mode" error. The first version is not valid $LaTeX$ because it tries to interpret _ as a subscript, which is illegal in text mode, so the underscore needs to be escaped if we want a literal underscore.
• Usage of \begin{aligned*}...\end{aligned*} within Sphinx's .. math:: directive without the :no-wrap: option, which leads to an "Erroneous nesting of equation structures" error. There are two ways to fix this: Either add the :no-wrap: option to the math directive to prevent Sphinx from erroneously wrapping it in another environment, or replace the align* environment with the aligned environment, which is designed to be used inside another equation environment.

These errors have always been present in our code, but I'm guessing they only showed up recently due to an update to MathJax, which renders the equations in web browsers. I'm not totally sure, but most likely older versions of MathJax were more forgiving of $LaTeX$ commands that didn't follow proper syntax.

The AI caught a few other instances of typos, bad formatting, and bad syntax that I also fixed, even though they weren't throwing errors.

I can't guarantee that I've caught all the errors, or that I've necessarily implemented the correct fix in all instances since the fixes were made programmatically via Python scripts from the AI. After I thought I had caught everything, I tried having the AI run a headless browser (Playwright) to scrape error messages from the rendered web pages (this idea was one of the AI's suggestions), and it caught one more that had been missed. So maybe that means now there shouldn't be any more errors?

[NathanielBlairStahn]

I had the AI search for missed errors again, this time by running MathJax directly, and it found a few more that escaped both the pattern matching in the initial Python script and the Playwright page scraping for some reason.

[NathanielBlairStahn]

For reference, here are a few of the pages that were affected by these changes:

• Azithromycin intervention
• Maternal hemorrhage
• Ischemic heart disease
• Measures of risk in epidemiology
• Atrial fibrillation and flutter
• Diarrheal diseases risk effects
• Dynamic wasting model (GBD 2020)

[zmbc]

With this fix:

• Azithromycin intervention
• Maternal hemorrhage
• Ischemic heart disease
• Measures of risk in epidemiology
• Atrial fibrillation and flutter
• Diarrheal diseases risk effects
• Dynamic wasting model (GBD 2020)

[zmbc]

I randomly sampled parts of this PR and they all looked reasonable

[SylLutze]

Similar to Zeb, I did a bit of random spot checking, I did not do a complete review of all lines.