ensure notebook-testing script fails when notebooks fail

[jameslamb]

Description

The notebook-testing CI job in this repo does not actually cause a loud CI failure if any errors are detected in notebooks. That's because of a bash mistake I made in #1407... that PR moved notebook-checking into a function, but didn't add a set -E to be sure errors from inside that function were appropriately trapped.

This PR fixes that:

• ensures that notebook failures actually cause CI failures
• fixes 2 typos in nyc_taxi_years_correlation.ipynb code
  • (not caught in Unskip taxi notebook from CI #1422 because of this CI script bug)

Context: #1422 (review)

Notes for Reviewers

How I tested this

Originally did not skip any notebooks. Saw the failures in one of the notebooks cause an actual merge-blocking CI failure.

Build link: https://github.com/rapidsai/cuspatial/actions/runs/10199784404/job/28219162698?pr=1424

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[jameslamb]

This is ready for review. The failing unit test jobs are unrelated... they're the result of #1427, which will hopefully be fixed by @mroeschke 's PR: #1428

[jakirkham]

Thanks James! 🙏

Looks good. Had one minor suggestion

[jakirkham]

There were a couple CI jobs that appear to have had issues starting up (this is well before anything cuSpatial related AFAICT)

Have gone ahead and restarted those failing jobs. Hopefully that clears out the issue

[jakirkham]

/merge

[jameslamb]

Thanks @jakirkham !

I think we'll need #1434 to be finished to unblock CI in this repo.

[jameslamb]

One the notebooks is failing with what looks like a new cudf-compatibility issue. Wrote that up here: #1437

[jakirkham]

Going to covert to draft for clarity while those other blockers get cleared out

[harrism]

I think draft should be reserved for PRs unready for review. This PR is already reviewed, but can't be merged. I suggest using the DO NOT MERGE label, which does what it says on the tin.

[jameslamb]

I think draft should be reserved for PRs unready for review.

No problem, I've moved this back to Ready for Review and added the DO NOT MERGE label.

[jameslamb]

Hopefully CI here will pass once #1438 is merged.

[jameslamb]

Thanks so much for all the help, everyone! Sorry this took a couple iterations, but I think this testing is in a better state now.