[simd.math] Fix various indexing problems

[eisenwave]

Fixes #9200

1. There was indexing of simd::riemann_zeta without any declaration in the code block below.
2. Indexing for simd::remquo was missing.
3. Various declaration blocks were overly large for out-of-line indexing. For consistency, all blocks declaring multiple functions were converted to inline indexing. This change is in the same style as [cmath.syn] Convert uses of \indexlibraryglobal to \libglobal #9163 If you look at the PDF, you will find that several of the code blocks in [simd.math] span multiple pages, so the library index ends up pointing to the wrong page, sometimes off by multiple pages.

[mattkretz]

Looks good from what I can see on my phone (on vacation).

[tkoeppe]

You're taking a vacation, in the moment where we need to index zeta functions?!

[tkoeppe]

@eisenwave could you run a visual diffpdf before/after?

[eisenwave]

@eisenwave could you run a visual diffpdf before/after?

no difference

[tkoeppe]

Wait, how can there be no difference if we were previously indexing something that didn't exist? Wouldn't that have shown up erroneously in the index?

[tkoeppe]

Oh, you've restricted the page numbers. Can you run the full diff, and show where the index starts to differ?

[eisenwave]

Just kidding; I need to rerun some stuff.

[eisenwave]

The first index difference is for simd::beta, which got shifted from 1928 to 1929.