[upgrades] Get cluster version: API server not responding - add retry

[przemyslavic]

We should add a retry to the following task

2019-11-19T19:53:36.1222091Z 19:53:36 INFO cli.engine.ansible.AnsibleCommand - TASK [upgrade : Get cluster version] *******************************************
2019-11-19T19:53:38.0041819Z 19:53:38 INFO cli.engine.ansible.AnsibleCommand - fatal: [vm-qa-upgrades-rhel-master-001]: FAILED! => {"changed": false, "cmd": "set -o pipefail && kubectl version --short -o json | jq --raw-output '.serverVersion.gitVersion'", "delta": "0:00:00.189137", "end": "2019-11-19 19:53:37.647744", "msg": "non-zero return code", "rc": 1, "start": "2019-11-19 19:53:37.458607", "stderr": "The connection to the server 10.0.0.6:6443 was refused - did you specify the right host or port?", "stderr_lines": ["The connection to the server 10.0.0.6:6443 was refused - did you specify the right host or port?"], "stdout": "null", "stdout_lines": ["null"]}