Decimal arithmetic dtypes and widening rescale cast

[gatesn]

Binary::return_dtype derives Hive-style result types for decimal + - * / (mirroring arrow-arith's execution-time rules, with precision saturation at the physical width), instead of erroring at plan time. The decimal cast kernel gains widening rescale so expression coercion can align decimal operand types.

[codspeed-hq]

Merging this PR will degrade performance by 11.63%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 2 improved benchmarks
❌ 5 regressed benchmarks
✅ 1521 untouched benchmarks
⏩ 10 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	chunked_bool_canonical_into[(1000, 10)]	20.7 µs	35.4 µs	-41.66%
❌	Simulation	chunked_varbinview_canonical_into[(1000, 10)]	161.9 µs	198 µs	-18.24%
❌	Simulation	chunked_varbinview_into_canonical[(1000, 10)]	177.1 µs	213.2 µs	-16.89%
❌	Simulation	decompress_rd[f64, (100000, 0.01)]	845.9 µs	980.8 µs	-13.75%
❌	Simulation	decompress_rd[f64, (100000, 0.1)]	845.9 µs	980.7 µs	-13.75%
⚡	Simulation	decompress_rd[f64, (100000, 0.0)]	1,024.6 µs	845 µs	+21.25%
⚡	Simulation	decompress_rd[f32, (100000, 0.0)]	586.8 µs	498.6 µs	+17.68%

Tip

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

---

_{Comparing ngates/decimal-arithmetic-dtypes (9257307) with develop (31167ca)}

Footnotes

1. 10 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[joseph-isaacs]

This looks very slow?