kubectl scale: Add dry-run prefix to indicate result is not applied

Currently, if user executes `kubectl scale --dry-run`, output has no
indicator showing that this is not applied in reality.

This PR adds dry run suffix to the output as well as more integration
tests to verify it.

staging/src/k8s.io/kubectl/pkg/cmd/scale/scale.go

@@ -144,16 +144,18 @@ func (o *ScaleOptions) Complete(f cmdutil.Factory, cmd *cobra.Command, args []st
 	if err != nil {
 		return err
 	}
+
+	o.dryRunStrategy, err = cmdutil.GetDryRunStrategy(cmd)
+	if err != nil {
+		return err
+	}
+	cmdutil.PrintFlagsWithDryRunStrategy(o.PrintFlags, o.dryRunStrategy)
 	printer, err := o.PrintFlags.ToPrinter()
 	if err != nil {
 		return err
 	}
 	o.PrintObj = printer.PrintObj
 
-	o.dryRunStrategy, err = cmdutil.GetDryRunStrategy(cmd)
-	if err != nil {
-		return err
-	}
 	dynamicClient, err := f.DynamicClient()
 	if err != nil {
 		return err
@@ -238,7 +240,10 @@ func (o *ScaleOptions) RunScale() error {
 	for _, info := range infos {
 		mapping := info.ResourceMapping()
 		if o.dryRunStrategy == cmdutil.DryRunClient {
-			return o.PrintObj(info.Object, o.Out)
+			if err := o.PrintObj(info.Object, o.Out); err != nil {
+				return err
+			}
+			continue
 		}
 
 		if err := o.scaler.Scale(info.Namespace, info.Name, uint(o.Replicas), precondition, retry, waitForReplicas, mapping.Resource, o.dryRunStrategy == cmdutil.DryRunServer); err != nil {