Remove data-field from RArray

[CGMossa]

To prepare for reviewing the other RArray PR, I decided to take a look and study this bit first..

I don't see any particular reason to have RArray equipped with a wild pointer.
I've made adjustments so that this is no longer needed / expected.

I'm more interested in feedback on this..... because it is me trying to understand what was solved here.

[JosiahParry]

@CGMossa I believe that this PR is failing because R-devel is 4.4 now and the windows machine cannot find rtools 44. Do you have the same assessment?

[CGMossa]

The issue is with this rextendr snippet:

    if (identical(R.version$crt, "ucrt")) {
      # TODO: update this when R 5.0 is released.
      if (!identical(R.version$major, "4")) {
        cli::cli_abort("rextendr currently supports R 4.x", class = "rextendr_error")
      }

      if (package_version(R.version$minor) >= "3.0") {
        rtools_version <- "43" # nolint: object_usage_linter
      } else {
        rtools_version <- "42" # nolint: object_usage_linter
      }

      rtools_home <- normalizePath(
        Sys.getenv(
          glue("RTOOLS{rtools_version}_HOME"),
          glue("C:\\rtools{rtools_version}")
        ),
        mustWork = TRUE
      )

I can confirm that the failing CI machine is a ucrt enabled one.

We might want to ping @Ilia-Kosenkov for this one.

[JosiahParry]

Thank you for confirming.

I am going to merge this PR as it simplifies and clarifies the RArray struct. No modifications to the tests had been made and all pass. The only user facing change is the removal of the data argument from RArray::from_parts() which is a trivial fix for users when they next upgrade—assuming they've done this at all.

We will resolve the rtools 44 issue in another PR as it is out of the scope of this PR