[Python] Interchange - test failing on MacOS builds

[AlenkaF]

Describe the bug, including details regarding any error messages, version, and platform.

Noticed tests started failing on MacOS CI builds, example:

=================================== FAILURES ===================================
___________________ test_pandas_assertion_error_large_string ___________________

    @pytest.mark.large_memory
    @pytest.mark.pandas
    def test_pandas_assertion_error_large_string():
        # Test AssertionError as pandas does not support "U" type strings
        if Version(pd.__version__) < Version("1.5.0"):
            pytest.skip("__dataframe__ added to pandas in 1.5.0")
    
        data = np.array([b'x'*1024]*(3*1024**2), dtype='object')  # 3GB bytes data
        arr = pa.array(data, type=pa.large_string())
        table = pa.table([arr], names=["large_string"])
    
        from pandas.api.interchange import (
            from_dataframe as pandas_from_dataframe
        )
    
>       with pytest.raises(AssertionError):
E       Failed: DID NOT RAISE <class 'AssertionError'>

/Library/Frameworks/Python.framework/Versions/3.11/lib/python3.11/site-packages/pyarrow/tests/interchange/test_conversion.py:294: Failed

from: https://github.com/apache/arrow/actions/runs/4917476625/jobs/8782656971

Component(s)

Python

[AlenkaF]

The AssertionError should start to fail with pandas 2.0.1 as the support for large strings was added to the interchange protocol implementation in pandas-dev/pandas#52795.

I do find it confusing that it errors only on MacOS. Plus, on my local machine (MacOS) it doesn't error (with 2.1.0 dev pandas version).
Edit: they are skipped due to the large_memory mark.