fix: check combined outputs for potential errors when getting bacalhau job id

[AquiGorka]

Review Type Requested (choose one):

• Glance - superficial check (from domain experts)
• Logic - thorough check (from everybody doing review)

Summary

These changes check the combined output from bacalhau create --id-only --wait path/to/job/spec for errors because I noticed that sometimes when the id was returned, the rest of the output also contained an error and the code was not checking for it, the rest of the process would execute but there would be no results in the output folder.

How to test this code? (optional)

I don't think I have a straightforward way to reproduce the error, I noticed the situation when testing this PR, I added a print statement to the runOutput var and I noticed the full output also included info to the lines of could not execute job, not enough nodes (I'm writing this from memory so if you are testing this bear in mind it may be a bit different) and the whole cli run was executed but the output folder did not include any files.
With these changes the process did return an error in that situation and the RP console showed the error had happened.

[walkah]

🚢 Looks good to me! If it's tricky to reproduce, perhaps worth a unit test to verify? (obviously that won't catch if bacalhau changes the output formatting...)

[AquiGorka]

If it's tricky to reproduce, perhaps worth a unit test to verify?

That's a great suggestion, tracking here because I need to focus on another task but I do want to get my hands dirty with unit tests - we have been relying on existing tests and have not written news tests ourselves.

obviously that won't catch if bacalhau changes the output formatting

We are pinning [email protected] for now because their next release will bring breaking changes.