skip tests with path issues on Windows

[Enchufa2]

The fix in 54ee8fd uncovered an issue with those two tests on Windows. Life is too short to debug path issues for tests on Windows, so this fix just skips them there.

Successful rhub tests here (there's one ERROR, but it's about vignette rebuilding, something about texi2pdf not being found): https://builder.r-hub.io/status/Rcpp_1.0.10.3.tar.gz-94d653c1f4064cf29639fc441187ce3d

Checklist

• Code compiles correctly
• R CMD check still passes all tests
• Prefereably, new tests were added which fail without the change
• Document the changes by file in ChangeLog

[kevinushey]

You could probably resolve the Windows issues by using system2() instead of system(), and using shQuote() on the file path passed in.

https://github.com/RcppCore/Rcpp/pull/1258/files?w=1#diff-4e4b42a31588c914e29a3de60cb92b14af71746ae9ab925744c967e8b51141c8R49

I'm also fine with just not worrying about it though, as long as we're certain it's an issue with the test itself and not with Rcpp (which appears to be the case).

[eddelbuettel]

Changeset is great

[eddelbuettel]

Happy to merge as is, question if we should try the suggestion by @kevinushey ? Or not bother?

[Enchufa2]

I think we shouldn't bother. Those tests already were "let's be super cautious" tests. They pass on Unixes, then everything is fine.

[eddelbuettel]

We'll know how r-universe fares in a bit. I'll report back.

[eddelbuettel]

Our CI failed on the very regular / unchanged 'update container' step. Has to do with Debian experimental changing its name to rc-buggy so for now I am just going to remove the apt list file.

[eddelbuettel]

That did the trick so we're all green at GH. Wonder if r-universe will pick up now

[eddelbuettel]

And in the meantime r-universe caught up and we are now four out of five:

The Apple 4.3 fail is my overeager switch to a higher precision qnorm comparison:

test_xptr.R...................    8 tests OK  17.5s
  ----- FAILED[data]: test_stats.R<[246](https://github.com/r-universe/rcppcore/actions/runs/4515895162/jobs/7953727465#step:7:252)--250>
   call| expect_equal(runit_qnorm_log(-1e+05)$lower, -447.197893678525)
   diff| Expected '-447.197893678525', got '-447.197494465048'
  Error: 1 out of 1590 tests failed

and I'll reverse that out and make sure we only compare to a lower tolerance of, say, 1e-6

> all.equal(-447.197893678525, -447.19749446504, tol=1e-5)
[1] TRUE
> all.equal(-447.197893678525, -447.19749446504, tol=1e-6)
[1] TRUE
> all.equal(-447.197893678525, -447.19749446504, tol=1e-7)
[1] "Mean relative difference: 8.927e-07"
>