Fix Provisioning of disconnected VolumeTemplate

Since provisioning from OpenStack Volumes was enabled, situation when original
Volume template can be disconnected from EMS by refresher can happen and it
doesn't look to be possible fix it on OpenStack side.

This PR allows EMS lookup fallback to options[:src_ems_id] when getting EMS
from source object fails.

Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1518381

app/models/manageiq/providers/openstack/cloud_manager/provision/cloning.rb

@@ -1,4 +1,13 @@
 module ManageIQ::Providers::Openstack::CloudManager::Provision::Cloning
+  def find_destination_in_vmdb(ems_ref)
+    super
+  rescue NoMethodError => ex
+    # TODO: this should not be needed after we update refresh to not disconnect VmOrTemplate from EMS
+    _log.debug("Unable to find Provison Source ExtmanagementSystem: #{ex}")
+    _log.debug("Trying use attribute src_ems_id=#{options[:src_ems_id].try(:first)} instead.")
+    vm_model_class.find_by(:ems_id => options[:src_ems_id].try(:first), :ems_ref => ems_ref)
+  end
+
   def do_clone_task_check(clone_task_ref)
     connection_options = {:tenant_name => options[:cloud_tenant][1]} if options[:cloud_tenant].kind_of?(Array)
     source.with_provider_connection(connection_options) do |openstack|