feat: add validation for reserved MPI environment variables

[VishalPainjane]

This PR implements validation logic in the MPI runtime plugin to prevent users from manually setting internal MPI environment variables in TrainJob.spec.trainer.env.

Currently, users can accidentally set environment variables like OMPI_MCA_orte_default_hostfile which conflicts with the operator's internal configuration. This can cause the MPI launcher to fail to discover workers correctly, leading to training failures that are difficult to debug.

Changes

1. Define the Forbidden List (pkg/constants/constants.go):

   • Added MPIReservedEnvNames constant containing the reserved MPI environment variables:
     • OMPI_MCA_orte_default_hostfile
     • OMPI_MCA_plm_rsh_args
     • OMPI_MCA_orte_keep_fqdn_hostnames
     • OMPI_MCA_orte_set_default_slots

2. Enforce the Rule (pkg/runtime/framework/plugins/mpi/mpi.go):

   • Updated the Validate function to check user-provided environment variables
   • Rejects job creation with a clear error message when reserved envs are detected
   • Follows the same pattern used in the Torch plugin for TorchRunReservedEnvNames

3. Unit Tests (pkg/runtime/framework/plugins/mpi/mpi_test.go):

   • Added test case for single reserved env detection
   • Added test case for multiple reserved envs detection
   • Added test case confirming non-reserved envs are still allowed

Which issue(s) this PR fixes

Fixes #3126

Checklist

• Code follows the existing patterns in the codebase (mirrors Torch plugin validation)
• Unit tests added and passing
• go vet passes
• go fmt passes
• Docs included if any changes are user facing

[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 tenzen-y 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