[Code-ve]

Summary

Fixes #1974.

The import of evidently fails with pandas 3.1.0 because infer_dtype_from_object, an internal pandas utility, is no longer available.

This change removes the dependency on that private API and introduces a local dtype-inference helper based on pandas.api.types.pandas_dtype and NumPy dtype handling. The existing TestColumnsType logic now uses this helper at both dtype-inference call sites.

Implementation

• Replaces the import of the removed pandas.core.dtypes.common.infer_dtype_from_object utility.
• Adds _infer_dtype_type to handle supported dtype representations, including NumPy types, pandas extension dtypes, and string aliases.
• Updates both expected and actual column-type inference paths to use the helper.

Validation

• Tests: 36 passed across the data-integrity and test-suite test modules.
• Diff validation: git diff --check passed.
• Lint and formatting: Ruff checks passed locally.

Scope

The change is limited to dtype inference in data_integrity_tests.py and does not intentionally alter the surrounding test behavior.

[Code-ve]

Hi @ptimizeroracle,

Just following up on this PR. The changes and relevant tests are ready from my side. Whenever someone gets a chance, I'd appreciate a review and any feedback.

Thank you!