This is a follow up from https://issues.redhat.com/browse/ACM-10473

Post https://issues.redhat.com/browse/ACM-8509 we have a new case scenario for how metrics collection resources are deployed in the hub and normal spokes with respective namespaces, watch logic, resources used or unused. 
The e2e tests under https://github.com/stolostron/multicluster-observability-operator  should accommodate the new use case and validate the functioning of metrics collection separately in the hub and spokes. 

Currently in the testcases there is only one managed cluster i.e. the local-cluster and the tests expect the metrics collection deployments to be in add-on observability namespace and expect to have a manifestwork and observability addon. We need to adapt the test cases to verify that these resources are not created for hub . We need to also launch a normal spoke/managed cluster and to run the test cases against and verify the functionality with the resources created in ocm-addon-observability namespace

New test cases:

• Ensure hub metrics collection i.e. endpoint operator, metrics-collector and related secrets and configmaps are in ocm-observability namespace
• No ocm-addon-observability namespace is created for hub
• No observability addon is created for hub
• No manifestwork is created for hub