[CI] Test: Switch test instance

[xyao-nv]

Summary

Short description of the change (max 50 chars)

Detailed description

• What was the reason for the change?
• What has been changed?
• What is the impact of this change?

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

This PR switches all CI job runner labels from [self-hosted, gpu] to [self-hosted, gpu-stg] across four workflow jobs (pre_commit, test, test_policy, build_and_push_image_post_merge) and removes the *lscpu anchor reference from the test_policy job.

Design Assessment

Design is sound. Straightforward infrastructure change — runner label swap to target a different self-hosted runner pool. No runtime code affected.

Findings

🟡 Warning: Removal of *lscpu step from test_policy — .github/workflows/ci.yml:185
The *lscpu YAML anchor reference was removed from test_policy but remains in other jobs. If this was intentional (e.g., gpu-stg runners don't need CPU topology logging), this is fine. If accidental, it could make debugging hardware differences on staging runners harder. Worth confirming this removal is intentional.

🔵 Suggestion: PR description is unfilled — The PR body still has the template placeholders. A brief note like "Switching CI to gpu-stg runner pool for testing" would help future readers understand the motivation (temporary test vs. permanent migration).

Test Coverage

No new tests needed — this is a CI infrastructure change with no runtime code impact.

CI Status

pre_commit job is currently pending on the new gpu-stg runners. Worth confirming it completes successfully before merge to validate that the staging runners are properly configured.

Verdict

Ship it — Clean, minimal CI config change. Just confirm the *lscpu removal is intentional and that CI passes on the new runners.