Bug Description

Originally reported here:

https://bugzilla.redhat.com/show_bug.cgi?id=1857298

Description of problem:
If DeploymentServerBlacklist is defined in a stack when we construct the ansible playbook execution for the scale down actions, we end up provising two --limits to the ansible playbook which causes the target scale tasks to be run against all nodes not in the DeploymentServerBlacklist instead of the nodes targeted to be removed.

42430 354597 0.0 1.8 5097428 4778888 ? R 15:15 0:00 /usr/libexec/platform-python /usr/bin/ansible-playbook-3 /var/lib/mistral/overcloud/scale_playbook.yaml --limit overcloud-fc640compute-32,overcloud-fc640compute-20,overcloud-fc640compute-118,overcloud-fc640compute-123,overcloud-fc640compute-165 --become --timeout 600 --inventory-file /var/lib/mistral/overcloud/tripleo-ansible-inventory.yaml --limit !overcloud-fc640compute-20:!overcloud-fc640compute-32:!overcloud-fc640compute-118:!overcloud-fc640compute-123:!overcloud-fc640compute-165

How reproducible:
Always if DeploymentServerBlacklist is defined

Steps to Reproduce:
1. deploy overcloud with multiple compute nodes
2. configure DeploymentServerBlacklist with one of the compute nodes
3. Scale down defined compute nodes

Actual results:
nova-compute service removed from all computes not in the DeploymentServerBlacklist

Expected results:
nova-compute should only be removed from the targeted nodes for removal

https://review.opendev.org/#/c/741271

This was already fixed byhttps://review.opendev.org/c/openstack/tripleo-common/+/741271 during Victoria cycle.