Add retries to kustomize_deploy tasks

We observe the following errors randomly occurring from
`kubernetes.core.k8s_info` module, typically just after
waiting for the openstack operators InstallPlan to be finished:
`AttributeError: ''NoneType'' object has no attribute ''status''`.

This may be because the OpenShift cluster gets overwhelmed
and sometimes the module receives the unexpected response,
so in `custom_condition()` in `plugins/module_utils/k8s/waiter.py`
there is no `status` field in `resource` [1].

As a mitigation, let's try retrying the task after some delay.

[1] https://github.com/ansible-collections/kubernetes.core/blob/main/plugins/module_utils/k8s/waiter.py#L86

Author: sdatko

roles/kustomize_deploy/tasks/install_operators.yml

@@ -166,6 +166,10 @@
           type: Ready
           status: "True"
         wait_timeout: 300
+      retries: 3
+      delay: 60
+      register: _cert_manager_operator_pods
+      until: _cert_manager_operator_pods is success
 
     - name: Wait for cainjector pods
       kubernetes.core.k8s_info:
@@ -179,6 +183,10 @@
           type: Ready
           status: "True"
         wait_timeout: 300
+      retries: 3
+      delay: 60
+      register: _cainjector_pods
+      until: _cainjector_pods is success
 
     - name: Wait for webhook pods
       kubernetes.core.k8s_info:
@@ -192,6 +200,10 @@
           type: Ready
           status: "True"
         wait_timeout: 300
+      retries: 3
+      delay: 60
+      register: _webhook_pods
+      until: _webhook_pods is success
 
     - name: Wait for certmanager pods
       kubernetes.core.k8s_info:
@@ -205,6 +217,10 @@
           type: Ready
           status: "True"
         wait_timeout: 300
+      retries: 3
+      delay: 60
+      register: _certmanager_pods
+      until: _certmanager_pods is success
 
     - name: Create catalog source and switch dependent operators to consume it
       when: