A few simple CSL fixes

[Omikhleia]

So as to make review easier, here are several fixes before I tackle with enhancements.¹

• fix(packages): Empty CSL macros must be suppressed = Mentioned in passing in Can't run either \printbibliography or \csl:reference.  #2202 (comment)

  Some macros are interpreted as present while they should apparently be considered as absent. An example is the "translator" macro, which generates a "Tradução " even in the absence of translator(s) afterwards. Re-reading the CSL 1.0.2 specification, however, the only place where content suppression is clearly mentioned is on cs:group (acting as implicit conditional), but there's no such element here... It's still unclear to me where the intended behavior is defined...

  The same entries have the macro suppressed by citeproc-js, and it sounds logical (though poorly documented).

• fix(packages): Invalid link on CSL DOI, PMID, PMCID with affixes = Also mentioned in the above-mentioned discussion:

  There's an interesting problem with links (with URL, DOI, etc.) and here again the CSL specification is somewhat obscure and I'd even say rather inconsistent (conflating usage for affixes...)

• fix(packages): Error handling the locator on some CSL styles = Not reported before, but MLA for instance was affected (and likely a bunch of others, we were just lucky Chicago styles have an explicit condition). Just some bad code ;)

• fix(packages): Correctly honor affixes on multiple CSL citations = Not reported before, but preparing the way to support citing multiple keys in our package, I realized the logic was wrong. The CSL specs are so unclear at places.

Footnotes

1. Note to @jodros: The two first fixes correspond to issues I saw after fixing an (unrelated) issue of yours. This being said, there are still some weird things with UFES-ABNT, UNEAL-ABNT or ABNT-IPEA, notably places where there is either repeated punctuations (e.g. Lorem. . Something) or missing spacing (e.g. (Some editors, EDS.)Some collection title). However, I see the same issues on these entries with that JavaScript-based online BiBTeX Converter, so I am unsure it's really our fault. ↩

[alerque]

I'm not yet sure why CI has blown up, but that's probably on me to fix.