Implement proper benchmarks with expectations

[StarpTech]

Hi, as titled. I have very good experience with https://github.com/k6io/k6 where you can define thresholds for different metrics like request duration, latency etc... Those tests should run in the CI and fail when the threshold can't hold.

For example: https://github.com/StarpTech/graphql-registry/blob/main/benchmark/composed-schema.js
CI: https://github.com/StarpTech/graphql-registry/blob/main/.github/workflows/bench.yml