Add retry for backup get_url task

As CentOS aria2 package is still only in epel-testing repo,
get_url tends to fail with timout errors, which forces to do rechecks
and creates extra load on gates.
So we're adding retries for this task and extending timeout.

Change-Id: I26d3c9338247ffc789bf4a5291e65e954151cc51

tasks/lxc_cache_preparation_systemd_new.yml

@@ -37,6 +37,11 @@
     dest: "/tmp/{{ cache_basename }}"
     validate_certs: "{{ (lxc_hosts_validate_certs | bool) | lower }}"
     force: true
+    timeout: 15
+  register: image_download
+  until: image_download is success
+  retries: 3
+  delay: 2
   when: job_result.failed
 
 - name: Ensure systemd-importd is enabled

tasks/lxc_cache_preparation_systemd_old.yml

@@ -46,6 +46,11 @@
     dest: "/tmp/{{ cache_basename }}"
     validate_certs: "{{ (lxc_hosts_validate_certs | bool) | lower }}"
     force: true
+    timeout: 15
+  register: image_download
+  until: image_download is success
+  retries: 3
+  delay: 2
   when: job_result.failed
 
 - name: Place container rootfs