Streamline fuzzer workflow

[robert3005]

• Separate build from fuzz run
• Minimise every corpus
• Checkpoint the corpus in the background so we don't lose progress on spot preemption

[codspeed-hq]

Merging this PR will degrade performance by 8.39%

⚠️ 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.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
❌ 3 regressed benchmarks
✅ 2194 untouched benchmarks
⏩ 206 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	arrow_checked_add_u32_neon[16384]	13.3 µs	20.4 µs	-34.43%
❌	Simulation	random_i8[0.5]	68 µs	91.6 µs	-25.77%
❌	WallTime	words_gather_scalar_avx2[65536]	8.3 µs	9.4 µs	-11.71%
⚡	Simulation	random_i16[0.8]	93.7 µs	75.4 µs	+24.38%
⚡	WallTime	arrow_checked_add_u32_avx2[16384]	21.3 µs	17.7 µs	+20.72%

Tip

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

---

_{Comparing rk/fuzzci (e25974b) with develop (3451462)}

Footnotes

1. 206 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. ↩