Fix worker count env parsing

[fallintoplace]

Summary

• Cast the NUM_WORKERS environment fallback to int in both wait_for_workers.py copies.
• Report an argparse error when NUM_WORKERS is set to a non-integer value.

Why

--num-expected-workers is parsed as an integer, but the environment fallback kept NUM_WORKERS as a string. When the CLI option was omitted and NUM_WORKERS was set, the worker-count check compared num_workers < num_expected_workers as int < str and raised TypeError.

Validation

• python3 -m py_compile scripts/dask/wait_for_workers.py mg_utils/wait_for_workers.py
• Smoke-tested both scripts with a temporary dask.distributed.Client stub and NUM_WORKERS=16.
• Smoke-tested both scripts with NUM_WORKERS=abc to confirm argparse exits with a clear error.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[ChuckHastings]

/okay to test 21ceb39

[ChuckHastings]

/ok to test b1fd420

[alexbarghi-nv]

👍

[alexbarghi-nv]

/merge