Strip out $$ as well

[blegat]

Rebases #1283 and strips $$ as well

Edit by @mortenpi: close #1283, fix #1278

[mortenpi]

Thanks for adopting that PR 😄 A few things to think about before we merge:

1. Will we ever be in a situation where we'll have these delimiters nested? E.g.. \[$$ ... $$\]? I feel that this will always be a bug in the generator of LaTeX.

   So, with that in mind, I am wondering if we should just if ... else match for the patterns and return as soon one matches, instead of passing the output to the next regex?

2. What are your thoughts on also stripping the $ ... $ and \( ... \)?

3. As was mentioned in Fix text/latex output in HTMLWriter #1283 (comment), it's not 100% clear what the correct thing to do is if there are no detectable delimiters. We could fall back to Utilities.mdparse, but that means that we require the show methods to print delimiters.. i.e. for example if show prints just x^2 then that won't be rendered as an equation. But this might be fine, since the MIME is text/latex, which is not specific to TeX's math mode.

4. It would be good to have examples of all the delimiters we handle in the example/test file as well.

[blegat]

I agree with 1. and 4, these are addressed in the new version of the PR

About 2. and 3., I agree with

But this might be fine, since the MIME is text/latex, which is not specific to TeX's math mode.

The new version of the PR is non-breaking for cases where it is not wrapped by brackets or $$ so this case may be left for later in case there is anything to be done.

[mortenpi]

Implementation LGTM, just the test examples need fixes. I think we'll merge this after 0.25.3 is tagged.

[mortenpi]

Thanks again for adopting and iterating on this @blegat!