fix: don't downcast large_string to string unnecessarily in concat_str for PyArrow

[MarcoGorelli]

closes #2097

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

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

Related issues

• Related issue #<issue number>
• Closes #<issue number>

Checklist

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

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

[MarcoGorelli]

this would break a plotly test

FAILED tests/test_optional/test_px/test_px_hover.py::test_sunburst_hoverdict_color[pyarrow] - pyarrow.lib.ArrowInvalid: Schema at index 1 was different: 
labels: large_string
parent: large_string
id: large_string
pop: double
country: large_string
continent: large_string
lifeExp: double
vs
labels: large_string
parent: string
id: large_string
pop: double
country: large_string
continent: large_string
lifeExp: double

I don't want to rush this then, leaving it out of tomorrow's release, we'll think about it for the next one

[MarcoGorelli]

The plotly test could be fixed by making nw.lit('a') default to large_string in PyArrow. But...should we?

will continue on the issue

[dangotbanned]

@MarcoGorelli I think I can shorten this and avoid the # type: ignore[arg-type].

Mind if I add a commit?

[MarcoGorelli]

sure thaknks

[dangotbanned]

@MarcoGorelli I think I can shorten this and avoid the # type: ignore[arg-type].

I managed to get there in fewer characters - but LOC is bound by the name chunked_arrays 😄

(4dfc6a3)

I'll conclude my round of code golf there for today

[dangotbanned]

This is more of a question than a suggestion.

Is there a preference between these two?

If not provided

and

If not specified

I probably wouldn't have noticed this if the lit(..., dtype=...) doc hadn't shown up in the diff

dtype: The data type of the literal value. If not provided, the data type will
be inferred by the native library.

[MarcoGorelli]

dunno, don't really mind

[MarcoGorelli]

thanks Dan!