🐛 Ensure move uses mutated metadata when updating a target object

[dlipovetsky]

What this PR does / why we need it:
The createTargetObject function is called as part of the move operation. It creates the target object, i.e. the object on the target cluster. If the target object already exists, the function updates the object.

Previously, the function fixed the name and namespace of the target object before it applied the experimental mutators. If some mutator modified the namespace of the target object, the function would continue to use the namespace established before the mutators were applied.

This adds a new (failing) unit test, and ensures that the target object namespace reflects the changes made by the experimental mutators.

--

We discovered this issue in a more complex scenario than the one in the new unit test.

In our scenario, we started the move operation, passing a mutator that changes the target namespace. The createTargetObject function try to create a ClusterClass in the target namespace. However, CAREN would copy the same ClusterClass to the namespace, and so the function would try to update it.

Because of the bug fixed here, the function read the ClusterClass in the default namespace on the target cluster, and then tried to update the ClusterClass in the target namespace, all the while using the UID and resource version of the ClusterClass in the default namespace:
https://github.com/mesosphere/cluster-api//blob/d6afeb8e33d9565680345b84e900c1d78a8ab76f/cmd/clusterctl/client/cluster/mover.go#L1005-L1019

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #

[takirala]

Thanks @dlipovetsky !

@cwyl02 we need to backport this to 1.6.x release cuz we cant consume 1.7.x in kommander due to controller runtime.