fix(framework): clean up Volcano PodGroup when TrainJob transitions to Failed

[Ayush-AM]

What this PR does / why we need it:
When a Volcano-scheduled TrainJob transitions to the Failed condition (or reaches a terminal state), its associated scheduling.volcano.sh/v1beta1 PodGroup object was never deleted or cleaned up. Consequently, orphaned PodGroups remained in the cluster indefinitely, reserving minResources in Volcano scheduler queues and starving pending jobs.

This PR addresses the issue cleanly and modularly:

1. Runtime Framework Extensibility: Introduced the TerminalCleanupPlugin interface (TerminalCleanup(ctx, trainJob) error) in pkg/runtime/framework so schedulers can clean up gang-scheduling resources on job completion without polluting the reconciler with vendor-specific logic.
2. Volcano Plugin Implementation: Implemented TerminalCleanup in the Volcano plugin (pkg/runtime/framework/plugins/volcano). It looks up the PodGroup corresponding to the TrainJob, validates ownership via metav1.IsControlledBy(podGroup, trainJob), and deletes it idempotently (client.IgnoreNotFound).
3. Controller Lifecycle Integration: Invoked r.Runtime.TerminalCleanup(ctx, &trainJob) in TrainJobReconciler.Reconcile() once trainjob.IsTrainJobFinished(&trainJob) is true.
4. RBAC Updates: Added delete verb for scheduling.volcano.sh/podgroups in RBAC role definitions (manifests/base/rbac/role.yaml and charts/kubeflow-trainer/templates/rbac/clusterrole.yaml).
5. Unit Tests: Added extensive unit tests across the Volcano plugin (TestVolcano_TerminalCleanup covering 6 test cases), Framework core (TestRunTerminalCleanupPlugins), and TrainJob reconciler (TestReconcile_TerminalCleanup verifying terminal cleanup on Failed and Complete jobs).

Which issue(s) this PR fixes:
Fixes #4033

Checklist:

• Docs included if any changes are user facing (internal controller gang-scheduling cleanup; no user-facing API schema changes)

/assign @kuizhiqing @robert-bell
/cc @kuizhiqing @robert-bell

[google-oss-prow]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign gaocegege for approval. For more information see the Kubernetes Code Review Process.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[github-actions]

🎉 Welcome to the Kubeflow Trainer! 🎉

Thanks for opening your first PR! We're happy to have you as part of our community 🚀

Here's what happens next:

• If you haven't already, please check out our Contributing Guide for repo-specific guidelines and the Kubeflow Contributor Guide for general community standards.
• Our team will review your PR soon! cc @kubeflow/kubeflow-trainer-team

Join the community:

• Slack: Join our #kubeflow-trainer Slack channel.
• Meetings: Attend the Kubeflow AutoML and Training Working Group bi-weekly meetings.

Feel free to ask questions in the comments if you need any help or clarification!
Thanks again for contributing to Kubeflow! 🙏

[Ayush-AM]

Hi @kuizhiqing @robert-bell! 👋

Submitted PR #4034 to fix issue #4033 ("Volcano PodGroup is not cleaned up when TrainJob transitions to Failed"):
#4034

All CI checks (PR title, DCO, and tests) are passing. Looking forward to your reviews! Thank you!