Add retries for removing datavol mountpoint if busy

There are some corner cases, like [1], when unmounting our (encrypted or not)
data volume succeeds but the subsequent mount remove task fails;
presumably this is a strange race condition due to the speed of
execution between the two tasks.

Make mount remove task more resilient but adding retries if it fails.

Closes elastic#47
Relates  elastic#48

[1] https://groups.google.com/a/elastic.co/d/msg/build-elasticsearch/nK8N3QQlpxI/1eE3WZBcAQAJ

Author: dliappis

fixtures/ansible/roles/encryption-at-rest/tasks/linux/main.yml

@@ -33,6 +33,10 @@
     dump: 0
     passno: 2
     state: absent
+  register: remove_mount_result
+  retries: 5
+  delay: 10
+  until: remove_mount_result | success
 
 - name: populate passphrase file for encrypted device
   copy:

fixtures/ansible/roles/initialize-data-disk/tasks/linux/main.yml

@@ -33,6 +33,10 @@
     dump: 0
     passno: 2
     state: absent
+  register: remove_mount_result
+  retries: 5
+  delay: 10
+  until: remove_mount_result | success
 
 - name: delete cryptsetup passphrase file if present
   file: