Replace all -F usage with -f

[jarmak-nv]

This pull request updates multiple GitHub Actions workflows to standardize the usage of the gh api graphql command by replacing the -F flag with the -f flag for all GraphQL variable inputs. This change ensures consistency and aligns with best practices for the gh CLI.

From GitHub

Using variables
In all of the following examples, you can use variables to simplify your scripts. Use -F to pass a variable that is a number, Boolean, or null. Use -f for other variables. For example,

my_org="octo-org"
my_num=5
gh api graphql -f query='
  query($organization: String! $number: Int!){
    organization(login: $organization){
      projectV2(number: $number) {
        id
      }
    }
  }' -f organization=$my_org -F number=$my_num

For more information, see Forming calls with GraphQL.

I'm unsure if this is a recent change, but we hit the failure when attempting to convert an integer into a string. GitHub may silently fix string-likes.

targeting python3.13 as cuDF found the failure; can merge to main after confirmation of resolution.

[jameslamb]

@jarmak-nv before we merge... I noticed this was targeted at the python-3.13 branch instead of branch-25.06... I suspect that's a mistake.

It does looks like there are some lingering references to the @python-3.13 branch (GitHub Search), but the PR for it (#268) has already been merged. I don't think we're planning to merge any new commits from the python-3.13 back into branch-25.06 at this point.

I've retargeted this at branch-25.06. I tried to do a rebase and force push, but I can't do that because this came from your fork. Could you please:

1. rebase this on the latest branch-25.06
2. force-push those changes to this branch

And then I think the deployment story should be:

1. merge this into branch-25.06
2. merge branch-25.06 into Add CUDA 12.9.0 #339
3. these changes will be out on all the repos branch-25.06 branches once the CUDA 12.9 migration is done and they're switched back to branch-25.06 (cc @gforsyth @bdice )

[vyasr]

I think there's some confusion in part because it's going to be a bit before all the repos are using the same branch of shared-workflows anyway at this point, and I'm the one asking Ben for these workflow fixes that are blocking cudf's board automations (I don't think any other repo is using them, or at least not to the same extent) but cudf is lagging in some of the migrations due to e.g. the CUDA 12.9 compiler issues. Let's stick with James's suggestion for the merge plan. Worst case we need to merge down branch-25.06 into one of the migration branch sooner for more testing.

[jarmak-nv]

Should be g2g

Will note cuML also uses these, currently pinned to branch-25.06.

[jarmak-nv]

@jameslamb - when you're on, any thoughts on this CI failure?

Lint GitHub Actions workflow files.......................................Failed
- hook id: actionlint-docker
- exit code: 1

Executable `docker` not found

Pre-commit linting passes on my end, but maybe I broke something? 😅

[gforsyth]

@jarmak-nv -- this looks like the recently added actionlint-docker check doesn't work on pre-commit.ci -- it's also broken on main, so I don't think you broke anything.

[bdice]

Sorry! I didn’t realize that hook required Docker. I just copied it from rapidsai/shared-actions. Perhaps we need to disable it in CI, or run in our own infrastructure.

[gforsyth]

I'll push up a commit disabling it for now -- I think we can run it on our own infrastructure but I need to familiarize myself with how we run workflows within shared-workflows

[gforsyth]

Ok, disabled it for CI only, so it'll still run locally. Going to merge this now

[gforsyth]

Thanks @jarmak-nv !

[jarmak-nv]

Thanks all! Appreciate the support 🚀