Honest benchmark callback

[DZakh]

I noticed that for some libraries, the operation is inlined instead of being called inside the benchmark callback. Since there are many libraries that could be potentially inlined like this eg Zod, but they still used the benchmark callback, I've decided to walk through all the libraries that don't use it and fix them. I think it'll make the benchmark more honest.

[moltar]

Thanks, @DZakh!

[hoeck]

I've read that function call overhead in V8 is basically nonexistent - but a consistent notation in the benchmarks is good to have too :+1: