Implement other definitions for success

[Mikulas]

Rollout status currently blocks until all requested pods are successfully started. This is intentional as we don't want to start QA and other parts of the CI/CD pipeline with parts with part of the traffic still being served from old pods. This should be added to documentation.

Furthermore, the rollout may fail for some pods (mounting volumes, resource constraints, …) even when a one pod is successful.

So far I don't see a use case for other modes, but I'm actively looking for them.