GH-137623: Use an AC decorator for docstring line length enforcement

[AA-Turner]

After thinking about it, @serhiy-storchaka's suggestion to use a decorator seems a better approach. It keeps the information localised to where the problem is, rather than hiding it in libclinic, which most contributors shouldn't need to worry about.

Opening as draft to avoid review-request spam until ready for merge.

A

• Issue: Argument Clinic: Enforce summary line length #137623

[AA-Turner]

I'd prefer to merge this, I think it's a cleaner solution than the current. @serhiy-storchaka do you have any objections?

A

[serhiy-storchaka]

ДПЕЬю Ж+1Ж

[serhiy-storchaka]

LGTM. 👍

[AA-Turner]

ДПЕЬю Ж+1Ж

Google translate doesn't have any suggestions here!

[AA-Turner]

Thank you!

[picnixz]

Oh, I would have suggested @allow_* instead of permit but it's merged now.

[AA-Turner]

Sorry, I wanted to minimise review spam so kept it in draft. Happy to change it if you'd like?

My rationale for "permit" instead of "allow" was to signal that this is an exception to the rule -- we don't want to see the number of exceptions go up here. "allow" could be seen as a supported way to bypass the checks, which we'd prefer to avoid.

I realise that this is a very slight semantic difference, though!

[picnixz]

Oh, if that's your rationale, I'm ok with it though in this case I would have had some "ignore_maxline_length" instead or "disable_docstring_length_check" but I'm fine the permit prefix.

[skirpichev]

Google translate doesn't have any suggestions here!

ДПЕЬ is LGTM in Russian (or Ukrainian, I think that part of keyboard has no difference). Shame on Google!