fix: poll job.status() once per job in Experiment.status()

[andrewwhitecdw]

Summary

Experiment.status() called job.status(runner=...) twice while building one status row: once for the display text and again for the returned dictionary. A backend failure therefore produced two identical ERROR tracebacks per status request and wasted a redundant status poll.

This was flagged as a non-blocking follow-up in #569.

Changes

• Store job.status(runner=self._runner) in a local variable and reuse it for both the display string and the returned dict in nemo_run/run/experiment.py.
• Add a regression test in test/run/test_experiment.py asserting job.status() is called exactly once for both return_dict=True and the default print path.

Testing

uv run -- pytest test/run/test_experiment.py::test_experiment_status_polls_job_once -xvs

Result: 1 passed.

Ruff lint and format also pass on the changed files.

Contributor guidelines

• DCO sign-off included.
• One fix per PR.
• Includes regression test.

[ko3n1g]

/review

[svcnemo-autobot]

🔎 On it — resolving the pull request and reviewing it rigorously.

[svcnemo-autobot]

🤖 On it — handling this as review.

[svcnemo-autobot]

💬 LGTM — #582

Verdict: No blocking findings.

Blocking findings: None.

Non-blocking findings: None.

Review details:
• nemo_run/run/experiment.py:911-929 polls job.status() once and consistently reuses that result for both the rendered status and returned dictionary, eliminating the duplicate backend request without changing the API or fallback behavior.
• test/run/test_experiment.py:659-680 covers both return_dict=True and the default display path, asserting exactly one status call with the experiment runner in each case.
• I reviewed both changed files, the full Experiment.status() implementation, Job.status()/JobGroup.status(), CLI and test callers, and current main. The intervening timestamp change on main is unrelated and produces no merge conflict.

Validation: All 18 non-skipped checks on the exact head f9e6067d2eee39a38363034c35686abec0afb6da pass, including the test matrix, lint, format, CodeQL, DCO, spell check, copyright, and secrets detection. git diff --check also passes. I could not independently run pytest because this sandbox has no Python runtime; CI provides the executable test coverage.