reconsider x hackery

[pkra]

We still have some hackery for the x element from the earliest days.

texml-to-html/lib/elements/x.js

Lines 23 to 24 in 964ecee

	// ignore if not xref/x or isBook
	if ('xref' !== xmlnode.parentNode.tagName && xmlnode.closest('article'))

(which will change slightly in #443)

Let's consider mapping x to some HTML element and let downstream delete those it doesn't like.

[pkra]

With texml generating more x elements, we need to move on this. In particular, statement titles will now wrap their parens in x elements - and we don't want to loose those in articles (cf. proc 16467)

[pkra]

Besides some lingering texml issues, I'm running into trouble dealing with x inside tex-math, i.e. ref/cite in math mode.

Right now, downstream journal content wants everything removed but the comma inside cite-detail. We could preserve enough markup but then our tex-math output will not be straight forward to use with MathJax (then again, it formally isn't right now either since we don't provide delimiters). Otherwise, we'd need to keep hacking x elements here (i.e., look up xml nodes).

So

1. expose cite-detail and x elements inside tex-math (and move a small bit further away from a mathjax-compatible output) -- then downstream adjusts those as needed
2. add tex-math specific logic for x elements, including our custom needs for journals

Right now I'm leaning towards option 1.