Array operations fuzzer operates on smaller arrays

[robert3005]

I believe some of the reference implementations are very slow so we can't
operate on bigger arrays

Signed-off-by: Robert Kruszewski github@robertk.io

[codspeed-hq]

Merging this PR will degrade performance by 39.88%

⚠️ 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.

⚡ 8 improved benchmarks
❌ 8 regressed benchmarks
✅ 1182 untouched benchmarks

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

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	mix[50%_in/50%_out]	340 µs	393.2 µs	-13.54%
⚡	WallTime	mix[100%_in/0%_out]	500.7 µs	449.5 µs	+11.39%
❌	WallTime	for[10M_u16]	72.3 µs	120.3 µs	-39.88%
❌	WallTime	for[10M_u32]	114.6 µs	164.5 µs	-30.34%
⚡	WallTime	dynamic_dispatch_u32[10M]	142.3 µs	94.2 µs	+51.09%
❌	WallTime	dynamic_dispatch_f32[10M]	114.7 µs	166.9 µs	-31.29%
⚡	Simulation	decompress_rd[f64, (10000, 0.01)]	138.6 µs	121.9 µs	+13.7%
❌	Simulation	decompress_rd[f32, (100000, 0.01)]	495.1 µs	582.7 µs	-15.04%
⚡	Simulation	decompress_rd[f64, (10000, 0.1)]	138.7 µs	122 µs	+13.63%
❌	Simulation	decompress_rd[f32, (100000, 0.1)]	495.1 µs	582.7 µs	-15.04%
⚡	Simulation	decompress_rd[f32, (10000, 0.0)]	94.6 µs	85.9 µs	+10.14%
❌	Simulation	decompress_rd[f64, (100000, 0.01)]	842.6 µs	1,020.5 µs	-17.44%
⚡	Simulation	decompress_rd[f32, (10000, 0.1)]	90.2 µs	82 µs	+10%
❌	Simulation	decompress_rd[f64, (100000, 0.1)]	842.5 µs	1,020.7 µs	-17.46%
⚡	Simulation	decompress_rd[f64, (10000, 0.0)]	138.5 µs	122.1 µs	+13.46%
⚡	Simulation	decompress_rd[f32, (100000, 0.0)]	583.5 µs	495.7 µs	+17.7%

---

_{Comparing rk/arrayopsfuzzer (964b3f1) with develop (cb9b138)}

[joseph-isaacs]

We can do this. But we do really want the 16k size going forwards

[robert3005]

I kept the file fuzzer working on 16k arrays since that looked fine. I think we need to spend some time on reference impls in the fuzzer