[Test automation] Fix pytest collection errors and schedule daily CI/CD workflow run

[ConnorHack]

What does this PR do?

This PR introduces a new daily cron job scheduled to run at 2:30 am EST daily. It also modifies the following small items:

• Fix an input reference.
• Remove ollama installation dependency, as it is no longer needed after the test framework refactoring.
• Update to v4 of GitHub Actions artifacts (link)

This PR also fixes some collection errors that were found when testing.

Test Plan

Workflow execution done within synced forked repository of llama-stack here.

Pre-commit job ran here.

[sixianyi0721]

hey @ConnorHack, from the log of this CI run, it seems tests were not passing (ex. "Run Tests: Loop"step) - how should we interpret the result? can you help me understand how the CI run result would be informative to developers?

[ConnorHack]

@sixianyi0721

hey @ConnorHack, from the log of this CI run, it seems tests were not passing (ex. "Run Tests: Loop"step) - how should we interpret the result? can you help me understand how the CI run result would be informative to developers?

If you're referring to the ChildFailedErrors: these are not actual errors. The (exit code: 5) failure occurs because no tests are collected, so this shouldn't be seen as tests not passing.

As for how to interpret the result...

...I need to take another look at what fixtures are enabled with the provider meta_reference and model llama_3b. Both test_model_registration.py and test_text_inference.py used to be collected and tested with these decorators, but this changed awhile ago ago and I need to investigate the reason.

So:

can you help me understand how the CI run result would be informative to developers?

TLDR: It isn't helpful in its current state, but it will be once I can identify which tests should be selected using the pytest parameters provider_id and llama_3b so there are tests both collected and selected.