Refactor internal helpers which generate fragmented messages

[MichaelChirico]

Noted in #6532. We should also address the group.desc issue, but in principle this PR could be submitted first. cc @aitap

[github-actions]

Generated via commit 93cfd19

Download link for the artifact containing the test results: ↓ atime-results.zip

Time taken to finish the standard R installation steps: 3 minutes and 30 seconds

Time taken to run atime::atime_pkg on the tests: 9 minutes and 21 seconds

[tdhock]

helper function duplicated_values looks ok.
I think we should remove group.desc argument, and add a new argument, called.from.measure default FALSE.

• If TRUE then we use stopf("... arguments to measure should be uniquely named, problems: %s")
• If FALSE then we use stopf("elements of fun.list should be uniquely named, problems: %s")
  or is there some better way of handling the translation issue?

[MichaelChirico]

I went with "just drop group.desc altogether", I think the error will be clear enough. measurev() is mentioned, which is documented next to measure() with the argument fun.list. If we really wanted to keep the distinction, I'd suggest looking at sys.calls() or something like that, rather than adding a new argument which is kind of useless when measurev() is exported.

[tdhock]

yes that is a good point that we shouldn't be adding arguments that the user can not provide

[MichaelChirico]

OK, I'll merge here. If we decide to decide internally that parent is ... and change the message, we can do so as a follow-up :)