Remove dataframe protocol

[vyasr]

Description

Follow-up to #17736

Checklist

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

[mroeschke]

Hmm darn. Ideally we should have had warnings as errors enabled for cudf pandas tests based on the configs in pyproject.toml, but because we're using a pytest.ini for these tests it looks like we didn't anticipate the test failure here

[vyasr]

Hmm yeah that is unfortunate. We should fix that in a follow-up PR. I don't know that we can reuse existing pyproject.toml files though given the special nature of these tests. Maybe our best bet is to just copy over filters from our other pyprojects?

[galipremsagar]

/merge

[galipremsagar]

I think we will need to investigate the massive increase in duration of pandas-tests job.

On a different pr it is(50 mins): https://github.com/rapidsai/cudf/actions/runs/13120888958/job/36623191308

On this pr it is consistently going beyond (1hr 30mins): https://github.com/rapidsai/cudf/actions/runs/13128280990/job/36670004745?pr=17909

[galipremsagar]

I think we will need to investigate the massive increase in duration of pandas-tests job.

On a different pr it is(50 mins): https://github.com/rapidsai/cudf/actions/runs/13120888958/job/36623191308

On this pr it is consistently going beyond (1hr 30mins): https://github.com/rapidsai/cudf/actions/runs/13128280990/job/36670004745?pr=17909

I think we will need to investigate the massive increase in duration of pandas-tests job.

On a different pr it is(50 mins): https://github.com/rapidsai/cudf/actions/runs/13120888958/job/36623191308

On this pr it is consistently going beyond (1hr 30mins): https://github.com/rapidsai/cudf/actions/runs/13128280990/job/36670004745?pr=17909

I'm investigating what is causing this slowdown.

[mroeschke]

Maybe not the root of the slowdown, but it appears tests/io/test_pickle.py::test_pickle_generalurl_read is taking a long amount of time on the 3 times pandas-tests was retried (the http calls are supposed to be mocked, but maybe not well enough)

2025-02-04T17:07:30.8282675Z [gw2] [ 99%] FAILED tests/io/test_pickle.py::test_pickle_generalurl_read[http://url.com] 
2025-02-04T17:36:11.9497790Z tests/io/test_pickle.py::test_pickle_generalurl_read[ftp://test.com] 

[vyasr]

Marked as do not merge until Prem has a chance to get to the bottom of the perf hit.