Fix string reductions and isin across string dtype flavors

[galipremsagar]

Description

Fixes several string-dtype behaviors to match pandas across all string dtype flavors (object, the four pd.StringDtype storage/na_value combinations, and pd.ArrowDtype string types), plus a cudf.pandas proxy fix. These resolve 21 xfailed pandas unit tests in the pandas-testing plugin.

• StringColumn.sum empty identity: summing an empty or all-null series now returns the additive identity — 0 for object dtype, "" for string dtypes — matching pandas (tests/arrays/string_/test_string.py::test_reduce_empty).
• min/max NA sentinel identity: _get_nan_for_dtype returns the dtype's exact na_value singleton for pd.StringDtype and arrow string dtypes, so min/max(skipna=False) satisfies result is dtype.na_value (pd.NA for "string" dtypes and ArrowDtype, the np.nan float singleton for "str" dtypes). Other kind-"O" extension dtypes (categorical, arrow decimal/binary) intentionally keep the float NaN fallback, since pandas coerces their skew/cov/corr results to a float NaN.
• isin across string dtype flavors: StringColumn._process_values_for_isin aligns string-typed values with the column's dtype (mirroring the numeric override), so isin matches on element values instead of returning all-False when the flavor differs.
• cudf.pandas object-ndarray identity: _transform_arg returns an object-dtype ndarray as-is when no element needed transforming. Rebuilding an equivalent copy broke aliasing checks such as np.may_share_memory(np.asarray(x), x), which numpy.random.Generator.permutation uses to decide whether to defensively copy before an in-place shuffle — under pandas copy-on-write this raised ValueError: array is read-only.

Test coverage added:

• Classic tests in tests/series/methods/test_reductions.py and tests/series/methods/test_isin.py, parametrized over object dtype, all four pd.StringDtype flavors, and pd.ArrowDtype(pa.string())/pa.large_string(), comparing against pandas wherever pandas does not raise or diverge (divergences are documented inline).
• cudf.pandas tests: a _transform_arg identity unit test and an end-to-end Generator.permutation test validated against real pandas with the same seed.

Checklist

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

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 67b835ac-d55b-4e24-bd97-6e39bc2ac79a

📥 Commits

Reviewing files that changed from the base of the PR and between 6a4d6ae and c581ad2.

📒 Files selected for processing (1)

• python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py

💤 Files with no reviewable changes (1)

• python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py

---

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes

  • Improved string-column sum, min, and max behavior for empty and null-containing data.
  • Fixed isin comparisons across different string dtypes and null values.
  • Preserved NumPy object-array identity when no conversion is needed.
  • Improved handling of missing-value sentinels across string and nullable dtypes.

• Tests

  • Added coverage for string reductions, isin, permutations, null handling, and object-array identity across supported storage formats.

Walkthrough

Updates nullable result handling, string reductions and isin dtype alignment, object-array proxy transformation, pandas compatibility mappings, and corresponding parametrized tests.

Changes

String dtype and proxy behavior alignment

Layer / File(s)	Summary
Nullable result contracts python/cudf/cudf/utils/dtypes.py, python/cudf/cudf/core/column/numerical_base.py, python/cudf/cudf/tests/series/methods/test_reductions.py	_get_nan_for_dtype now returns typed nullable sentinels for string and extension dtypes, numerical early returns no longer suppress the return type, and string reduction tests cover empty, null, and object-dtype cases.
String reduction and membership behavior python/cudf/cudf/core/column/string.py, python/cudf/cudf/tests/series/methods/test_isin.py	Empty string reductions use dtype-specific identities, object-string isin values are cast to the left-hand dtype, and string dtype/null-handling tests cover these behaviors.
Proxy array identity preservation python/cudf/cudf/pandas/fast_slow_proxy.py, python/cudf/cudf_pandas_tests/test_fast_slow_proxy.py, python/cudf/cudf_pandas_tests/test_cudf_pandas.py	Object arrays are returned unchanged when transformations preserve element identity, rebuilt when proxy elements require conversion, and validated through string permutation tests.
Pandas compatibility mappings python/cudf/cudf/pandas/scripts/pandas-testing-plugin.py	Known-failure entries are removed for categorical replacement, string operations, permutation, reductions, and string copy-view cases.

Estimated code review effort: 3 (Moderate) | ~25 minutes

Suggested labels: improvement

Suggested reviewers: wence-, mroeschke, brandon-b-miller, Matt711

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main changes to string reductions and isin behavior across string dtype flavors.
Description check	✅ Passed	The description is directly aligned with the PR's string-dtype fixes, proxy behavior change, and added tests.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[galipremsagar]

/okay to test 6a4d6ae

[galipremsagar]

/okay to test c581ad2

[galipremsagar]

/merge