[bug] kcm controller doesn't create release and crashes if createManagement is set to false

[Algeran]

Describe the bug
Deployed kcm-controller v0.0.7 with disabled default Management object creation. Controller pod fails to start, prints in logs a lot of errors and doesn't create the release object.

To Reproduce
Steps to reproduce the behavior:

1. Values:

   controller:
     createManagement: false
   

2. deploy kcm helm chart
3. monitor logs and pod status, check that release object is not created

Expected behavior
Controller should start and create release kcm-0-0-7 release object

Additional context

controller.log

[kylewuolle]

I've tried with the latest code and this is now fixed:

1. git checkout 5c0ce3213dfc2c989e9ee4efeb266d0ee34fb21b (0.0.7 tag)
2. make kind-deploy registry-deploy dev-push
3. helm install kcm oci://127.0.0.1:5001/charts/kcm --version 0.0.7 -n kcm-system --create-namespace --set="controller.createManagement=false" --set="image.repository=localhost/kcm/controller"
4. I get the same error as in the issue
5. make dev-destroy
6. git checkout c486cd1fdcc5e9af4d68afe7c74ccce836459c5c (current main revision)
7. make kind-deploy registry-deploy dev-push
8. helm install kcm oci://127.0.0.1:5001/charts/kcm --version 0.0.7 -n kcm-system --create-namespace --set="controller.createManagement=false" --set="image.repository=localhost/kcm/controller"
9. It now works as expected

[Algeran]

Hi @kylewuolle ,
I started the controller from the main revision as you described. Controller is now not crushing, but it still doesn't create release object. So, it's not possible to create any custom Management object due to it requires to specify the release name. In operator pod logs I can see a lot of errors from all internal controllers that they can't find the Management object to backup. And I think this is the reason why the release object is not created as well:

kcm/internal/controller/release_controller.go

Line 76 in c486cd1

err = r.Get(ctx, client.ObjectKey{Name: kcm.ManagementName}, management)

This condition blocks reconciliation loop in all controllers. Moreover,

1. it's not clear why it tries to backup the management object with the kcm name. I understand, that this is the default name for the Management object that is created automatically, but I can create the custom one with any name, not kcm. And it will block the reconciliation process for all CRDs that managed by the operator.
2. Why the Release reconciliation loop includes the Management object backup logic. As far as I understand, the relation between release and management objects is inverse - you can't create and reconcile management object without fully deployed release.

[Algeran]

Added the log file from the controller pod

controller.log

[zerospiel]

Not related actually, there is bit another kind of the problem, related either to out local dev ops or how to properly set the latest version of te kcm-templates chart

I believe the recent changes from #906 are related to the errors above

The aforementioned LOC is not related though