fix: add maxUnavailable for cloud provider upgrade

Changing the value to 1 from the default 25% remedies the deployment
update stuck problem in single-node clusters due to pod antiaffinity.

Signed-off-by: Zespre Chang <zespre.chang@suse.com>

charts/harvester-cloud-provider/templates/deployment.yaml

@@ -11,6 +11,10 @@ spec:
   selector:
     matchLabels:
   {{- include "harvester-cloud-provider.selectorLabels" . | nindent 6 }}
+  strategy:
+    rollingUpdate:
+      maxUnavailable: 1
+    type: RollingUpdate
   template:
     metadata:
       labels: