Raise ValueError consistently for negative int -> unsigned conversions

[teddytennant]

Extracting a negative Python int into an unsigned Rust integer previously raised an inconsistent exception type depending on the width and the Python version: OverflowError for u8/u16/u32 (try_from after PyLong_AsLong) and u64 (PyLong_AsUnsignedLongLong), but ValueError for u128 on 3.13+.

This standardizes on ValueError for the negative case across every unsigned width (u8–u128, usize, and the NonZeroU* variants), matching the convention of newer CPython APIs such as PyLong_AsUInt64, as suggested in the issue. Genuine magnitude overflow of an in-range-sign value continues to raise OverflowError.

The remap is gated on <$rust_type>::MIN == 0 (and, in the 128-bit path, the existing $is_signed), and only fires on the already-cleared error path after confirming the value is negative, so signed types and too-large positive values are unaffected.

Adds the test_negative case to test_common! so it runs for all integer widths (signed types still accept -1; unsigned types must raise ValueError).

cargo test -p pyo3 --lib conversions::std::num

Fixes #6116.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[codspeed-hq]

Merging this PR will degrade performance by 12.26%

❌ 5 regressed benchmarks
✅ 135 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	iter_tuple	5.1 ms	6 ms	-14.89%
❌	tuple_get_borrowed_item_unchecked	2.9 ms	3.3 ms	-13.45%
❌	tuple_get_borrowed_item	3.5 ms	4 ms	-11.19%
❌	tuple_get_item_unchecked	3.6 ms	4.1 ms	-10.89%
❌	list_get_item_unchecked	3.7 ms	4.1 ms	-10.82%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing teddytennant:fix-negative-unsigned-valueerror (fddb340) with main (8fcf8fc)}