[R-package] Use R standard routines to access numeric and integer array data in C++

[jameslamb]

Another step towards #3016.

Fixes #4216.

Changes in this PR:

• removes LightGBM-custom pointers to R data (R_REAL_PTR(), R_INT_PTR()) with the standard equivalents from Rinternals.h (REAL(), INTEGER())
• converts arguments on the C++ side that are accessed that way from LGBM_SE to SEXP

Description

R is a dynamically-typed language. You can run code like x <- c(1, 2, 3) without declaring that x is a numeric array, and R will just figure it out.

R makes this possible by storing data for an object in a structure called a SEXPREC. That structure will contain different data based on the R class (integer, character, function, etc.). Libraries can reference that data in C code using a type, SEXP, provided by Rinternals.h. Libraries can also get a pointer to a particular type of data within a SEXPREC using other functions provided by Rinternals.h, for example REAL() to get a pointer to its numeric data.

As described in #4216, the internal details of the SEXPREC struct can change between different versions of R. SEXP is the official R API into that struct, and by using it libraries can reliably stay compatible with multiple R versions.

This PR's changes fix a bug that currently exists when using {lightgbm} with R 3.6, and protects {lightgbm} from similar bugs in the future.

For more details on this, see https://cran.r-project.org/doc/manuals/r-release/R-ints.html#SEXPs.

Notes for Reviewers

This PR does not depend on #4242 or #4247

[jameslamb]

Thanks to @david-cortes for providing the write-up in #4216, which included the reproducible example that inspired this unit test.

I tested this unit test tonight on my Mac, with R 3.6 and {lightgbm} 3.2.1 installed from CRAN, and confirmed that it failed in exactly the way described in #4216. So I'm confident that this test passing means the changes in this PR fix #4216.

[jameslamb]

/gha run r-solaris

Workflow Solaris CRAN check has been triggered! 🚀
https://github.com/microsoft/LightGBM/actions/runs/803646649

solaris-x86-patched: https://builder.r-hub.io/status/lightgbm_3.2.1.99.tar.gz-c0227bf93bb84b58be157cf6186f1119
solaris-x86-patched-ods: https://builder.r-hub.io/status/lightgbm_3.2.1.99.tar.gz-d66322e8d51b41f3a298625253478da5
Reports also have been sent to LightGBM public e-mail: http://www.yopmail.com/lightgbm_rhub_checks
Status: success ✔️.

[jameslamb]

/gha run r-valgrind

Workflow R valgrind tests has been triggered! 🚀
https://github.com/microsoft/LightGBM/actions/runs/803647172

Status: success ✔️.

[jameslamb]

/gha run r-valgrind

Workflow R valgrind tests has been triggered! 🚀
https://github.com/microsoft/LightGBM/actions/runs/807454258

Status: success ✔️.

[jameslamb]

/gha run r-solaris

Workflow Solaris CRAN check has been triggered! 🚀
https://github.com/microsoft/LightGBM/actions/runs/807454445

solaris-x86-patched: https://builder.r-hub.io/status/lightgbm_3.2.1.99.tar.gz-3701ffb1819540bf98d77218c211cd5c
solaris-x86-patched-ods: https://builder.r-hub.io/status/lightgbm_3.2.1.99.tar.gz-481aa16dd0084b4c9bae6fc8a0043cd7
Reports also have been sent to LightGBM public e-mail: http://www.yopmail.com/lightgbm_rhub_checks
Status: success ✔️.

[StrikerRUS]

Thank you very much for the iterative improvement!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.