Markdown filter: exclude MathML content

[Chase Tingley]

Apparently MathML in Markdown is supported by some tools, as I have samples from a user that do this.

The filter currently handles this via HTML code matching, which produces extreme tag garbage. We should just treat everything <mathml>...</mathml> as non-translatable.

[Chase Tingley]

Attached testcase.

I don't think flexmark handles mathml well, so this one may be a bit of a challenge.

[Kuro Kurosaka]

The original file had some unbalanced tags within math elements. This file fixes it and between <math> and </math> are now well balanced.

[Kuro Kurosaka]

The code changes in PR 212 should implement this feature. This, however, does not support math element or its parent (enclosing) HTML element written in a line. This is due to the limitation of flexmark-java library that the markdown filter uses. I have experimented and made a feature request of flexmark-java (https://github.com/vsch/flexmark-java/issues/193). If they accept it and implement it, we can change small changes to the markdown filter code to use it.

[Chase Tingley]

Markdown filter - replace codefinder handling of HTML with subfilter

Fix issue #645 (and improve #651) by using real subfiltering instead
of the codefinder.  Cleanup to quite a bit of test code.

→ <<cset 44b31f7719b5>>

[Chase Tingley]

Code is merged; I'm in favor of closing this. We can open a new issue for the remaining case.