Right now, the Connect builds are using the same operation timeout as all other operations. But in some cases, the Kafka Connect builds can take a long time - especially if they download lot of connectors or when they have slow connection for downloading artifacts / pushing images. Currently, when the build fails to finish within the timeout, it is deleted and a new build is started in the next reconciliation. So a user with a slow build has only one option - increase the operation timeout. But that increases the timeout for all operations, so it is not optimal.

The PR #4196 adds a new timeout dedicated to the Kafka Connect builds. But in addition to that, in a long term, we should also change the build process to not start new build every reconciliation but follow on the build left over from previous reconciliation.

This should be fairly straight forward for Kubernetes builds. But it might be a bit trickier for OpenShift builds, where we trigger the build and don't create the build directly, so we need to be able to find the last build etc.

Created by Strimzi#4195.