Description of problem:
If a node has had Ceph on it before and you deploy a fresh Ceph node over it, the deploy fails. To remedy this, currently we need to manually boot each node and "ceph-disk zap /dev/sdFOO" the disks before the deploy can succeed again. 

Version-Release number of selected component (if applicable):
Director GA w/ 0day

How reproducible:
Every time

Steps to Reproduce:
1. Deploy Overcloud w/ 1 or more ceph nodes
2. Delete the stack
3. Deploy again with the same configuration, reusing the hardware from the last deploy

Actual results:
Deploy fails to create ceph nodes.

Expected results:
Deploy succeeds, overwriting the old ceph disks. 

Additional info:
If doing this by default is too much of a change from existing behavior, consider adding a ForceDiskZap parameter to storage-environment.yml that will zap and re-read the partition table when provisioning the nodes. 

This has affected PoC deploys at several customer sites.