Add perf profiling to vitest

[GuillaumeLagrange]

No description provided.

[codspeed-hq]

CodSpeed Instrumentation Performance Report

Merging #55 will degrade performances by 19.02%

_{Comparing cod-1253-support-perf-profiling-with-vitest (453da47) with main (e31ee8b)}

Summary

⚡ 1 improvement
❌ 1 regression
✅ 80 untouched

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

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	test_iterative_fibo_10	423 µs	305.4 µs	+38.51%
❌	wait 500ms	10 ms	12.4 ms	-19.02%

[codspeed-hq]

CodSpeed WallTime Performance Report

Merging #55 will improve performances by 23.08%

_{Comparing cod-1253-support-perf-profiling-with-vitest (453da47) with main (e31ee8b)}

Summary

⚡ 6 improvements
✅ 42 untouched

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	test_iterative_fibo_10	420 ns	372 ns	+12.9%
⚡	test_iterative_fibo_100	720 ns	636 ns	+13.21%
⚡	test_recursive_cached_fibo_10	1.3 µs	1.1 µs	+11.7%
⚡	test sync baz 10	384 ns	312 ns	+23.08%
⚡	test sync baz 100	492 ns	444 ns	+10.81%
⚡	short body	1.9 µs	1.7 µs	+11.35%

[art049]

looks good, minor changes are left.

How do we distinguish between a sync and async bench here?

[GuillaumeLagrange]

How do we distinguish between a sync and async bench here?

@art049 there's no distinction currenlty because vitest does not make the distinction and runs everything using tinybench's async run entrypoint
Empirically, I've had much better results with vitest's stack traces than tinybench's. I have not narrowed down why though.