extend available_linters(tags = ), add linters_with_tags()

[AshesITR]

Fixes #980

I moved the with_defaults() test from test-settings.R to test-with.R and moved the implementation from zzz.R to with.R.

with_tags() also allows us to test that default_linters is synchronized to available_linters(tags = "default") so I did that.

[AshesITR]

Maybe adding a crossref from available_linters() to with_tags() is a good idea?
And maybe also from ?linters to help discover the new feature.
Since with_tags() can be used as a drop-in replacement for with_defaults() as long as default = is missing, I think with_tags() could even be documented to become the standard way of configuring lintr. WDYT?

[AshesITR]

Not sure what is up with codecov() erring out for all.equal() when ubuntu-release doesn't and both run R 4.1.3...

[AshesITR]

2022-03-28T20:45:06.8497300Z ##[error]Error: Failure in `/tmp/RtmpjHIBqn/R_LIBS2488420bd9ed/lintr/lintr-tests/testthat.Rout.fail`
2022-03-28T20:45:06.8506140Z ol_linter\": target, current do not match when deparsed"
2022-03-28T20:45:06.8506879Z - "Component \"trailing_blank_lines_linter\": target, current do not match when deparsed"
2022-03-28T20:45:06.8507591Z - "Component \"trailing_whitespace_linter\": target, current do not match when deparsed"
2022-03-28T20:45:06.8508204Z - "Component \"vector_logic_linter\": target, current do not match when deparsed"
2022-03-28T20:45:06.8508735Z + "Component \"line_length_linter\": Component \"length\": Mean relative difference: 0.5"

Sounds like the all.equal() check in codecov compares deparse(fun)...

Locally irreproducible:

f1 <- with_tags()[["vector_logic_linter"]]
f2 <- with_defaults()[["vector_logic_linter"]]
all.equal(f1, f2)
#> [1] TRUE
all.equal(deparse(f1), deparse(f2))
#> [1] TRUE

EDIT: Found the problem. covr::report() modifies functions to count line hits. These modifications cause inconsistent deparse() results in our case it seems.

[MichaelChirico]

EDIT: Found the problem. covr::report() modifies functions to count line hits. These modifications cause inconsistent deparse() results in our case it seems.

seems like a covr bug... will that block submission here?

[AshesITR]

Nope. Added a skip nevertheless.

[MichaelChirico]

Getting closer, thanks!

[AshesITR]

PR fails seem to originate elsewhere.
Also, the test warnings() are back it seems...

https://github.com/r-lib/lintr/runs/5749209064?check_suite_focus=true#step:6:177

EDIT: 🤦🏻 calling the internal function build_linter was the culprit.