Fix rollout race condition and display events in Jenkins log

[michaelsauter]

Fixes #382.

If there is an error while starting a rollout, we check if the version has advanced, and simply do not start a rollout.

Further, this PR introduces a new feature which I believe will help users a lot to debug issues. When something goes wrong with a deployment the cause can usually be found in either the events of the replication controller (some error prevents pod creation) or the events of the pod (some error prevents the pod from becoming active). Unfortunately, many people don't know where to look or don't have the rights to do so. To improve this situation, we now query for the events and display them in the Jenkins log when something prevents a successful rollout. It may look like this:

WARN: Error: hudson.AbortException: Deployment timed out. Observed related event messages:
=== Events from ReplicationController 'foo-20' ===
Created pod: foo-20-b6wmt
=== Events from Pod 'foo-20-b6wmt' ===
Successfully assigned myproject/foo-20-b6wmt to ip-172-31-52-140.eu-east-1.compute.internal
pulling image "172.30.21.193:5000/myproject/foo@sha256:ca88cd1ff2572cee6b74f7efdc5afc6f6d649a1e1616394a80da18dbe9601f4c"
Successfully pulled image "172.30.21.193:5000/myproject foo@sha256: ca88cd1ff2572cee6b74f7efdc5afc6f6d649a1e1616394a80da18dbe9601f4c"
Created container
Started container
Readiness probe failed: rpc error: code = 2 desc = oci runtime error: exec failed: container_linux.go:235: starting container process caused "exec: \"exit 1\": executable file not found in $PATH"

I have also reduced the code duplication between component and orchestration pipeline. I have done a basic test of the orchestration pipeline, but am planning to do more extensive testing anyway as I am working on #367 which requires a fair amount of refactoring.