lint x %in% NA in equals_na_linter

[MichaelChirico]

Closes #2088

In writing this I had three follow-up questions

1. Should we customize the lint message for %in%/==/!= (yes)
2. Should we lint x %in% c(NA, y) to encourage is.na(x) | x %in% y (I think no)
3. Should we lint NA %in% x (should be anyNA(x); I am less sure, maybe it could go under any_is_na_linter()?)

[codecov-commenter]

Codecov Report

Merging #2112 (7bfc14f) into main (99fdd85) will increase coverage by 0.00%.
The diff coverage is 100.00%.

❗ Current head 7bfc14f differs from pull request most recent head e252ee2. Consider uploading reports for the commit e252ee2 to get more accurate results

@@           Coverage Diff           @@
##             main    #2112   +/-   ##
=======================================
  Coverage   99.65%   99.65%           
=======================================
  Files         120      120           
  Lines        5240     5242    +2     
=======================================
+ Hits         5222     5224    +2     
  Misses         18       18           

Files Changed	Coverage Δ
R/equals_na_linter.R	100.00% <100.00%> (ø)

📢 Have feedback on the report? Share it here.

[AshesITR]

Re 3.: I Agree it should be linted in the any_is_na_linter()

[AshesITR]

Also should test that NA %in% x doesn't lint here.

[MichaelChirico]

Follow-ups:

#2113
#2114

[MichaelChirico]

Also should test that NA %in% x doesn't lint here.

Good call -- the logic was broken (==/!= throw lint for NA on LHS or LHS of the operator, not so for %in%). Done now