Never lower an R version requirement in use_data()

[jennybc]

Closes #2077

[olivroy]

@jennybc, the condition should be added in use_standalone() as well. This is the first place I encountered this

also in use_test(). if you depend on testthat 3.2.1 and call use_testthat(3), you don't want to decrease your testthat dep to 3.0.0?

[jennybc]

@olivroy Do you remember which standalone file caused the minimum R version to be downgraded for you?

Another option I was considering is a specialized use_package() helper for R itself, i.e. splitting that out so that use_package() can just focus on normal packages. I'll explore that solution as well.

[jennybc]

use_test() only calls use_testthat_impl() if the package doesn't use testthat yet. So that doesn't seem like it would really contribute to this problem. I'm not saying you didn't experience it with some specific set of moves, but in the main use case, I don't see the connection there.

[olivroy]

Ok! simple repro case..

If I have in DESCRIPTION,

Imports:
  rlang (>= 1.1.4)

With dev usethis,
I call usethis::use_standalone("r-lib/rlang", "types-check"), which requires rlang 1.1.0.

It will decrease the dependency of rlang silently.

[jennybc]

I'm just going to worry about not downgrading the required version of R. I think use_standalone() is enough of an expert-use only function that said expert can also decide not to stage other dependency changes.