cephadm doesn't reuse osd_id of 'destroyed' osds
3 - minor
Pull request ID:
The replacement operation is supposed to work like this:
ceph orch osd rm $id --replace
This will leave the osd.$id in the crushmap with the "destroyed" flag.
The next osd deployment should pick those up and pass it to ceph-volume
ceph-volume lvm batch <devices> --osd-ids <ids_to_reuse>