If the REST service is down or not responding (eg. because it's a cluster replica), there should still be a way to teardown that manager. Currently teardown always runs a request that isn't strictly required.
After and CFY-7231, this is no longer needed.
The way to teardown a node removed from a cluster is now:
cfy cluster nodes remove node_name
cfy profiles use 184.108.40.206 --skip-credentials-validation
cfy teardown -f --ignore-deployments