Set continue on error in the cudf-polars-rapidsmpf nightly CI job

[Matt711]

Description

Justification: The new rapidsmpf runtime is in flux, so nighlies can break frequently.

Checklist

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

[rjzamora]

Weird - I could have sworn continue-on-error: true was already there when this was first added. Sorry about that!

[TomAugspurger]

The error at https://github.com/rapidsai/cudf/actions/runs/19175022030/job/54820199137?pr=20550#step:12:60454 is fixed on main. I think this branch is just a bit behind (https://github.com/Matt711/cudf/blob/8022c2e4c82bdb1a7eadbc96f3ef03bb4a86abfa/python/cudf_polars/cudf_polars/experimental/rapidsmpf/utils.py#L95).

That job might be non-blocking though.

[Matt711]

The error at https://github.com/rapidsai/cudf/actions/runs/19175022030/job/54820199137?pr=20550#step:12:60454 is fixed on main. I think this branch is just a bit behind (https://github.com/Matt711/cudf/blob/8022c2e4c82bdb1a7eadbc96f3ef03bb4a86abfa/python/cudf_polars/cudf_polars/experimental/rapidsmpf/utils.py#L95).

That job might be non-blocking though.

Thanks, the nightly CI check was blocking. So I request an admin merge

[rjzamora]

It's strange the the cudf-polars + rapidsmpf tests were failing here. That CI job was using an old version of cudf-polars for some reason?