Invert the relationship between the operator and the helm chart

[chrisseto]

As mentioned in #91, the existing architecture requires us to continuously maintain two different codebases. To rectify that, we'll have to invert the relationship between the operator and the helm chart.

Concretely: This ticket can be closed once the helm chart deploys the operator, a redpanda CRD, and any window dressing that doesn't belong within the operator.

We'll phase this out in a backwards compatible way where the operator and a CRD is first deploy in a "noop" mode. From there, we'll slowly transfer the "ownership" of core resources (Statefulset and the like) into the operator. By utilizing the helm chart's go code, we should be able to seamlessly deploy the resource the same way.

JIRA Link: K8S-199