Attempt to address Coordinated Omission: wrk2...?

[bbros-dev]

At least we should acknowledge the issue - might be a suitable topic for a blog post; determine the effect on these benchmarks.

• wrk2: Both addresses CO and can be used to benchmark - as opposed to load test.
• autocannon: CPU saturation on the system running the benchmark is the risk.
• Locust Primarily distributed and gives little weight to this use case.
• vegeta: Currently more suited to load testing more than bench-marking - per this comment in issue #486. This may change when Integrate with FastHTTP (massive performance gains) tsenart/vegeta#486 is resolved (Integrate with fasthttp tsenart/vegeta#500)?
• fortio: Currently have not resolved their issue with coordinated ommision. But it is on their radar per check for measurement bias/error fortio/fortio#3