[ci] [R-package] allow more possibly-lost warnings from valgrind

[jameslamb]

Contributes to the v4.2.0 release (#6191).

The latest valgrind build showed 1,056 bytes "possibly lost", all from code paths that create threads w/ pthread_create. (#6191 (comment))

In the past, CRAN has allowed some of these as false positives. (#6191 (comment))

This PR proposes modifying the test script for that job to acknowledge that this set of findings from valgrind are allowed.

Notes for Reviewers

It would be better to suppress specific cases instead of do text-parsing to enforce a maximum number of "possibly lost" reported bytes... but that will take a bit more time to develop. I've documented it in #6231 and put good first issue on it to try to increase the likelihood that someone comes and contributes that.

[jameslamb]

/gha run r-valgrind

Workflow R valgrind tests has been triggered! 🚀
https://github.com/microsoft/LightGBM/actions/runs/7190693469

Status: success ✔️.

[jameslamb]

Could I please get a review on this this week? This is the last PR I think needs to be done before we put up release v4.2.0 (#6191). I'd like to get that release out this week, to get that quantized training fix (#5994) out to more users.

[jmoralez]

Approving to unblock, hopefully someone else reviews this.

[jameslamb]

Thanks @jmoralez ! I'm going to merge this to continue with testing the release candidate.

For what it's worth, the R package has been up for 11 days now and CRAN has not reported any valgrind issues yet.

https://cran.r-project.org/web/checks/check_results_lightgbm.html

[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.