Also lint nrow(filter(.)) in nrow_subset_linter()

[MichaelChirico]

Closes #2313

This is the simplest version. I was weighing a more complicated version a la conjunct_test_linter()'s allow_filter= argument where we also check that this is not stats::filter() and decided against it for now. FWIW, there are no hits for nrow(stats::filter( on GitHub (which of course doesn't preclude this being run without namespace-qualifying stats):

https://github.com/search?q=lang%3AR+%2Fnrow%5B%28%5D%5Cs*stats%3A%3Afilter%5B%28%5D%2F&type=code

I also left out new okay examples because AFAIK the with() approach is the simplest in {dplyr} world as well. The more {dplyr}-esque approach I can think of would be x |> summarize(n = sum(<filter>, na.rm = TRUE)) |> pull(n), but that looks way worse to me than with() (OTOH, it will work with all backends).

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (7cb5023) 98.54% compared to head (a8ca6fb) 98.54%.

❗ Current head a8ca6fb differs from pull request most recent head b33346d. Consider uploading reports for the commit b33346d to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2457   +/-   ##
=======================================
  Coverage   98.54%   98.54%           
=======================================
  Files         126      126           
  Lines        5720     5720           
=======================================
  Hits         5637     5637           
  Misses         83       83           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.