pronouns: fetch list from the pronoun.is GitHub repo at start

[dgw]

Description

We don't have to wait until the 12th of Never for pronoun.is to have an API. Parsing tab-separated values is easy, and the file is already sorted according to approximate frequency for us according to their readme.

Includes a kill-switch config setting in case the repo ever gets deleted or reorganized, or the file format changes.

Checklist

• I have read CONTRIBUTING.md
• I can and do license this contribution under the EFLv2
• No issues are reported by make qa (runs make quality and make test)
• I have tested the functionality of the things this change touches

Notes

Speaking of something that's over-complicated for what it does (as Exi said on IRC yesterday)… this also finds the shortest unique set of pronouns for each option to use in links. The only outlier cases now are they/them/their/theirs/themsel(ves|f), which won't be abbreviated as they/.../themsel(ves|f). I might or might not find the motivation to fix that.

[Exirel]

I don't really get the tree aspect of the pronoun parsing. I feel like I'm missing the problem that required a solution.

I also tried to run the code on the current pronoun file, and the result is quite different from the default set of pronouns. For example, instead of 'they/.../themselves', the parser returns 'they/them/their/theirs/themselves'. So, is it compatible? Why use ... in the default value, knowing that the actual file don't contain them? Especially given that the file will replace the default value, so if there is a feature here, it'll be made inaccessible.

A lot of confusion!

[dgw]

I don't really get the tree aspect of the pronoun parsing. I feel like I'm missing the problem that required a solution.

Using a trie was to make it easier to match on prefixes. I think. It's been a few months. 😅

I've tried (😏) to address most of the style comments in fixup commits. Haven't gone through another round of testing just yet.

[Exirel]

I still think this is a lot of work that I don't know how to justify. On the other hand, I didn't have to do anything, and it all looks good. Maybe I'm just annoyed there isn't any test for that yet, but there wasn't before anyway.

Let's move on.

[dgw]

I have cleaned the history (squashing fixups) and fixed one last bug with the disambiguation output (in case the user specifies e.g. .setpronouns they, which matches multiple sets).

Testing also determined that it's not necessary to merge the fetched pronouns with the default set; they/.../themself (for example) works just fine without that entry existing in the dictionary. I dropped that change.