Fix inconsistent gate and lint failures for os-swift

Gate was failing due to a race condition whereby the partition is not ready
and so the first mounted partitions fails to mount - causing hosts to fail.

Adding a retry/delay on failures in the mount task should resolve this issue.

ansible-lint failures were occurring due to a recent release of ansible-lint
Since the gate failure cannot be merged without the linting fix, this change
also contains I0cc973867d9c279eb6ea05decc5ebccfaa19c7b0 to break the co-
dependency between this and the other patch.

Change-Id: I3f4daaed764c4cad8a07408fedf8d0cba570a344

test-requirements.txt

@@ -1,4 +1,4 @@
-ansible-lint
+ansible-lint<=2.3.9
 ansible>=1.9.1,<2.0.0
 bashate
 flake8

tests/test.yml

@@ -258,10 +258,10 @@
       - 'swift1'
       - 'swift2'
     delegate_to: "{{ physical_host }}"
-  - name: Create the Swift mount points, fstab entries and mount the file systems
+  - name: Create the Swift fstab entries and mount the file systems
     mount:
       name: "/srv/{{ container_name }}_{{ item }}"
-      src: "/opt/{{container_name}}_{{ item }}.img"
+      src: "/opt/{{ container_name }}_{{ item }}.img"
       fstype: xfs
       opts: 'loop,noatime,nodiratime,nobarrier,logbufs=8'
       passno: 0
@@ -270,6 +270,10 @@
     with_items:
       - 'swift1'
       - 'swift2'
+    register: mount_status
+    until: mount_status | success
+    retries: 5
+    delay: 2
     delegate_to: "{{ physical_host }}"
   - name: Swift extra lxc config
     lxc_container: