Don't lint nested string literals on LHS of assignment

[MichaelChirico]

Closes #1466

There's a few different ways we could fix this. I think it would also be enough to look for nearby SYMBOL_FUNCTION_CALL.

I also think we can skip the LBB test for STR_CONST, but didn't bother looking too carefully for this PR.

NB: relevant test cases like {a,"a"}${b,"b"} <- 1 are already present in the suite:

lintr/tests/testthat/test-object_name_linter.R

Lines 150 to 155 in 3e6b862

	expect_lint("good_name$\"badName\" <- 42", NULL, linter)
	expect_lint("good_name$'badName' <- 42", NULL, linter)
	expect_lint("badName$\"good_name\" <- 42", lint_msg, linter)
	expect_lint("badName$'good_name' <- 42", lint_msg, linter)
	expect_lint("`badName`$\"good_name\" <- 42", lint_msg, linter)
	expect_lint("`badName`$'good_name' <- 42", lint_msg, linter)

[codecov-commenter]

Codecov Report

Merging #2038 (ade6443) into main (239de20) will not change coverage.
The diff coverage is n/a.

❗ Current head ade6443 differs from pull request most recent head 56126ef. Consider uploading reports for the commit 56126ef to get more accurate results

@@           Coverage Diff           @@
##             main    #2038   +/-   ##
=======================================
  Coverage   98.56%   98.56%           
=======================================
  Files         114      114           
  Lines        5013     5013           
=======================================
  Hits         4941     4941           
  Misses         72       72           

Files Changed	Coverage Δ
R/object_name_linter.R	98.18% <ø> (ø)