conjunct_test_linter: Allow to opt out of linting dplyr::filter()

[salim-b]

#2077 added support to lint logical conjunctions inside dplyr::filter() calls. Quote:

Relatedly, dplyr::filter(DF, A & B) is the same as dplyr::filter(DF, A, B), but the latter will be more readable / easier to format for long conditions.

I respectfully disagree with the "latter will be more readable" part and would like to use conjunct_test_linter without linting dplyr::filter() calls. Would you be willing to introduce a flag like allow_inside_dplyr_filter to opt out of this behaviour?

[MichaelChirico]

I think in that sentence the emphasis is on "for long conditions". We often see code like:

DF |>
  filter(
    is_treatment &
      (expenditure - cost > 0) &
      group == "x"
  )

In this case it's not very controversial that using , is better:

DF |>
  filter(
    is_treatment,
    expenditure - cost > 0,
    group == "x"
  )

It's more of a toss-up for simple cases, like A & B vs A, B.

Anyway, happy to accept a PR. I would simply call such an argument allow_filter.

[salim-b]

It's more of a toss-up for simple cases, like A & B vs A, B.

Yeah, but I also think & is clearer than , in differentiating logical conjunctions from logical disjunctions (|).

Anyway, happy to accept a PR. I would simply call such an argument allow_filter.

Cool, see #2110.