feat: validate duplicate column names in pyarrow and duckdb

[IsaiasGutierrezCruz]

closes #1920

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

• Related issue enh: validate duplicate column names in pyarrow and duckdb #1770
• Closes enh: validate duplicate column names in pyarrow and duckdb #1770

Checklist

• Code follows style guide (ruff)
• Tests added
• [] Documented the changes

If you have comments or can explain your changes, please do so below

To test the validations, I added three methods in the file tests/frame/schema_test.py. @MarcoGorelli, if you think these would be more appropriate in another file, please feel free to let me know. Additionally, I had to add duckdb as a dependency to the extreme GitHub Actions workflow to accommodate these tests.

[MarcoGorelli]

Thanks! I'll check the perf impact here in case it's significant

[FBruzzesi]

Thanks! I'll check the perf impact here in case it's significant

Thinking about this (also in the pandas case), we could add a flag to skip_columns_validation as not every operation from which we return self._from_native_frame(...) might lead to duplicate columns - i.e. select, with_columns, rename can generate duplicate column names, while other ops would not.

[MarcoGorelli]

yes, we may need from_native_frame and from_native_frame_unchecked

[MarcoGorelli]

thanks @IsaiasGutierrezCruz ! i just added a commit to elide the checks in some cases

[IsaiasGutierrezCruz]

@MarcoGorelli Thanks for taking care of this PR! Appreciate you getting it over the finish line.
I saw in the comments that cudf does not raise the DuplicateError. I will take a look