Test replica scale-down warning

https://github.com/longhorn/longhorn/issues/4120

Steps

Given Replica Auto Balance set to least-effort or best-effort.

And Volume with 3 replicas created.

And Volume attached to node-1.

And Monitor node-1 manager pod events.

kubectl alpha events -n longhorn-system pod <node-1 manager pod> -w

When Update replica count to 1.

Then Should see Normal replice delete event.

Normal    Delete                   Engine/t1-e-6a846a7a                                Removed unknown replica tcp://10.42.2.94:10000 from engine

And Should not see Warning unknown replica detect event.

Warning   Faulted                  Engine/t1-e-6a846a7a                                   Detected replica UNKNOWN-tcp://10.42.1.98:10000 (10.42.1.98:10000) in error
[Edit]