CLN/TST: normalize test_frame_apply

[rhshadrach]

• Ensure all linting tests pass, see here for how to run them

Some minor normalizations before breaking up or parametrizing these tests

[jbrockmendel]

@jorisvandenbossche there's an ArrayManager test currently failing on master; any obvious candidates for something recently-merged that might account for that?

[rhshadrach]

@jbrockmendel @jorisvandenbossche - Just looking at recent commits, the earliest one I see failing is #40050. This failed in the PR as well. Both as well as this PR have a single test failure, test_groupby_dtype_inference_empty.

[jorisvandenbossche]

Yes, that PR was simply merged with failing tests .. Merging master will resolve it now thanks to Brock

[jorisvandenbossche]

I personally don't find this an improvement. I know we are inconsistent about it, but if there is a full link, it's much easier to go to the issue

[jreback]

this is how we do for every other comment about the issue number

if we want to require a full link it's possible but would need a precommit hook

[jorisvandenbossche]

We don't do that for "every other comment". We currently use the full link in 400 cases.

Not every detail needs to be controlled with a pre-commit hook

[jreback]

and how many w/o a link? i don't know but i would say it's way higher

we have to have a standard and then enforce it

we simply cannot track stylistic things any other way or involve personal preference here (once we agree on a standard)

[rhshadrach]

@jorisvandenbossche - Thanks for the feedback here, I didn't realize there wasn't a consensus. I actually agree on the preference for a link (but also more strongly prefer consistency). I've reverted these changes.

Assuming we can get a consensus, if things were made consistent and we updated the dev docs, it seems to me that consistency would become the norm and not require enforcement after a short amount of time. If it doesn't turn out to be the case and becomes a hassle, we could then implement a pre-commit hook.