Benchmarks

[mrazauskas]

Few ideas regarding benchmarks (very good to see them, of course!).

1. What if the same assertion library would be used with all runners? Any of them, but the same. Might be node:assert, or chai, or anything else. For instance, Mocha can be also used with expect, but it is slower than node:assert. I have a feeling that benchmark here should test speed of a test runner, not speed of assertion library.
2. Consider using hyperfine instead of spawning a process.

[wellwelwel]

I'll take a look at hyperfine. I like the idea of testing the test runner independently, but I was curious why not include the assertions as well.

For more transparency, the current tests were inspired by the uvu benchmark.

[wellwelwel]

Although Poku uses assert and strict from Node.js, it performs a parser that varies according to the platform, both before and after the assertion result, including a minimal builder:

poku/src/modules/essentials/assert.ts

Line 6 in 5e69de5

export const assert = createAssert(nodeAssert);

• That's why I believe it's also interesting to use assert and ensure that Poku set is well balanced.