add tests for validating `helper.py` CLI interactions

[DavidKorczynski]

We recently had a regression where the reproducer command stopped providing the output from the container run, which meant that reproducer commands just stopped silently. We should improve such that we have tests that are a bit more rigid in terms of enforcing the CLI interaction we expect from helper.py.

Refs:

• regression: fix(helper): Remove PyYAML dependency to resolve CI failures #14145
• fix 1: infra/helper: pass the right arguments to docker_run in reproduce_impl #14178
• fix 2: infra: Fix reproduce not outputting Docker logs #14299

[pitrou]

I can help review such a PR if desired.