* Add statefulSet override

- new crd properties: spec.override.statefulSet, which is
our own definition of appsv1.StatefulSet, which allows users to provide
a patch that will be applied on top of generated statefulset definition
as an override.
- new type EmbeddedObjectMeta, which is a partially metav1.ObjectMeta;
it is used in StatefulSet, PodTemplateSpec, and PersistentVolumeClaim.
- custom types are needed because of the pruning behavior of
`preserveUnknownFields: false`, which is going to be the default
behavior for crd v1.
- all fields in the custom StatefulSetSpec are optional; including
selector, serviceName...etc

Declare Go 1.13 in module file

* Add environment setup to KinD README

Added a step to set up environment variables before running `make
deploy-kind`

* Fixed various typos in README

This brings spelling of resources to be more consistent:
* `RabbitmqCluster` for the CR
* `cluster-operator` for the operator & image

* Add supported version/License/Copyright to README

* helm charts cleanup
* Add mtls, remove repo references to charts makefile
* Update template test version number
* remove references in makefile
* Update wording
Co-authored-by: James McNeil <jmcneil@pivotal.io>

Add beta notice to README

Multiline description and mTLS property

This software is provided as a beta pre-release. This commit updates the
README to reflect this.

Remove GH actions cause we dont have access to them

* Remove CI-related code from repo

- CI specified config is moved to CI repo and rest is templated as env vars
- Rename OPERATOR_IMAGE and split with registry url
- Adds Makefil env var description to README
Co-authored-by: Feroz Jilla <fjilla@pivotal.io>

* a few rabbitmqcluster examples

* typo

Set default rabbit version to 3.8.5

- related to issue #167

Adds mutual TLS
Co-authored-by: James McNeil <jmcneil@vmware.com>

ReconcileSuccess status condition

Update Operator Chart metadata

We actively develop and build using Go 1.13. Our mod file should reflect
this information.

MPL license in controller and config files

Integration tests started to error as a side effect to update the CR
Status more frequently; those are fixed by adding a retry logic to
Update calls in tests.

- Moved the time transition logic to SetCondition
- ReconcilableCondition() is intended to be used as a initialiser
- Log CR Name and Namespace when there is an error
- Update unit tests to expect Reconcilable condition
- Rename Reconciable condition to Reconciled

Expose a condition to inform whether the custom resource can be
reconciled. We consider the custom resource to be reconcilable if we are
able to send a create or update request to KubeAPI and receive a success
response.

This is a building block that the controller can use to create and
expose a condition of type Reconcilable. Intended to indicate whether
RabbitmqCluster resource can be reconciled.

Rabbitmq operator system test in envs without LoadBalancer capabilities

Log stderr and stdout when unable to enable plugins

Co-authored-by: Aitor Cedres <acedres@pivotal.io>

- Some minor refactor to not use PivNet registry image in every test spec
- PivNet registry image is used to test the ability to access private
registries. It is enough to test this in only one spec
- Ginkgo only needs one hyphen for arguments
Co-authored-by: Feroz Jilla <fjilla@pivotal.io>

Node Port service type is externally accessible via the Kubernetes node
IP. We can fetch this IP from the Kubernetes API. The port that
redirects to RabbitMQ pods via the service is the NodePort in the
Service Spec. The requirement now is that system tests can access
Kubernetes node IP; it is slightly similar to the previous assumption,
where the system tests relied on the IaaS to provide an external
accessible IP.

The intention of SynchronizedBeforeSuite is to initialise singletons
e.g. a database. It then provides a block to initialise clients, using
the data from the previous block if needed. Since we don't have a
singleton, we don't need to use this.

Increase system tests timeout and reduce parellel nodes

- Previous timeouts were enough for majority of the time;
however roughly 1 out of 5 times, some test cases will need
much longer time to create pods and wait until pods are ready.
Pods do eventually become ready, and tests will pass