Fix manila-share with ceph-nfs backend

Currently when you deploy manila-share with a ceph-nfs backend
we get the following error:
Error: pcs -f
/var/lib/pacemaker/cib/puppet-cib-backup20210107-48280-1y4rwmx
constraint colocation add openstack-manila-share with ceph-nfs INFINITY
failed: Error: Resource 'ceph-nfs' does not exist. Too many tries

The reason for it is that we moved the pcs resource creation to the host
and outside containers
https://review.opendev.org/q/topic:%22pcs_host3%22+(status:open%20OR%20status:merged)

But when the ceph-nfs pcmk resource gets created at step 5, the
manila-share puppet code has already been invoked and since it requires
the ceph-nfs resource to be available it fails like above.

Let's move the ceph-nfs creation to step4 to solve this.

Closes-Bug: #1910732

Change-Id: Ifb81d155cbb4598a311c8e6fe90be951a8e7e656

manifests/profile/pacemaker/ceph_nfs.pp

@@ -91,7 +91,10 @@ class tripleo::profile::pacemaker::ceph_nfs (
     }
   }
 
-  if $step >= 5 and $pacemaker_master {
+  # When we create manila-share resource at step 5 we need the ceph-nfs pcmk resource up
+  # and running. But since we moved to pcs commands invoked on host, manila-share at step5
+  # gets created *before* ceph-nfs (as it is invoked via step_config vs docker_config)
+  if $step >= 4 and $pacemaker_master {
     pacemaker::resource::service { 'ceph-nfs' :
       service_name  => 'ceph-nfs@pacemaker',
       op_params     => 'start timeout=200s stop timeout=200s',