Good concept.
But I wonder how well such test tooling simulates/replicates cloud scale/performance related issues. This could be useful if you work out and anticipate cloud issues from latency, misbehaving services, down services, etc. that you can then attempt to simulate/inject.
But it might be hard if you don’t know what you might exactly encounter in production at some given load or scale of deployment. Not sure how you can mimic full production setup in this kind of case.