chore: clean up cond cuda compilation

[0ax1]

Always include CUDA code in tests and benchmarks if the environment has a CUDA installation. This does not require the user to pass a feature flag but is determined at build.rs time.

[codspeed-hq]

CodSpeed Performance Report

Merging this PR will degrade performance by 15.94%

_{Comparing ad/cleanup-cuda-cond-compilation (daca0eb) with develop (0b4b579)}

⚠️ 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.

Summary

⚡ 10 improved benchmarks
❌ 3 regressed benchmarks
✅ 1241 untouched benchmarks
🆕 20 new benchmarks
⏩ 1274 skipped benchmarks¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
🆕	WallTime	u16_FoR[1K]	N/A	5.6 µs	N/A
🆕	WallTime	u8_FoR[10M]	N/A	5.6 µs	N/A
🆕	WallTime	u32_FoR[100K]	N/A	7.3 µs	N/A
🆕	WallTime	u16_FoR[1M]	N/A	6.1 µs	N/A
🆕	WallTime	u8_FoR[1K]	N/A	7.5 µs	N/A
🆕	WallTime	u32_FoR[10K]	N/A	6.3 µs	N/A
🆕	WallTime	u8_FoR[1M]	N/A	5.6 µs	N/A
🆕	WallTime	u64_FoR[100K]	N/A	12.4 µs	N/A
🆕	WallTime	u16_FoR[10M]	N/A	9.7 µs	N/A
🆕	WallTime	u32_FoR[1K]	N/A	6 µs	N/A
🆕	WallTime	u64_FoR[10K]	N/A	14.3 µs	N/A
🆕	WallTime	u64_FoR[1K]	N/A	6.9 µs	N/A
🆕	WallTime	u64_FoR[10M]	N/A	342.5 µs	N/A
🆕	WallTime	u16_FoR[100K]	N/A	8.1 µs	N/A
🆕	WallTime	u32_FoR[10M]	N/A	174.2 µs	N/A
🆕	WallTime	u32_FoR[1M]	N/A	11.6 µs	N/A
🆕	WallTime	u8_FoR[100K]	N/A	5.6 µs	N/A
🆕	WallTime	u8_FoR[10K]	N/A	5.6 µs	N/A
🆕	WallTime	u16_FoR[10K]	N/A	6.3 µs	N/A
🆕	WallTime	u64_FoR[1M]	N/A	32.7 µs	N/A
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Footnotes

1. 1274 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. ↩