Cluster AutoScaler does not scale Machines when target is outside of min or max range

Issue

The Cluster Node Autoscaler does not scale if the current node count is outside of the maximum node-count or minimum node-count values for the autoscaler.

The below log is an example of the Autoscaler failing to scale down when the maximum node-count is set to 10, the target node-count is set to 5 and the current node-count is 16.

$ oc logs cluster-autoscaler-default-XXXX -n openshift-machine-api | grep "Problem with empty node deletion"
E1124 16:00:20.913422       1 scale_down.go:1147] Problem with empty node deletion: failed to delete worker-XXXX: size increase too large - desired:15 max:10