Add explicit cluster_url with oc login

[jwmatthews]

This PR fixes the issue I've hit recently when trying to deploy latest 3.7 origin builds with our ec2 scripts which replace the cluster cert with letsencrypt signed certificates. I think the replacement of the CA may be causing issues.

I changed these variables in my_vars.yml
origin_image_tag: latest
openshift_client_version: latest

./run_setup_environment.sh
...

...

TASK [openshift_setup : Login as admin] *******************************************************************************
fatal: [34.233.221.91]: FAILED! => {"changed": true, "cmd": "/usr/bin/oc login -u admin -p admin", "delta": "0:00:00.177283", "end": "2017-10-01 13:56:55.907736", "failed": true, "rc": 1, "start": "2017-10-01 13:56:55.730453", "stderr": "error: x509: certificate signed by unknown authority", "stderr_lines": ["error: x509: certificate signed by unknown authority"], "stdout": "", "stdout_lines": []}

If I log into the instance I see same error trying to execute "oc login":

[ec2-user@ip-10-0-0-70 ~]$ sudo su -
[root@ip-10-0-0-70 ~]# oc login
error: x509: certificate signed by unknown authority

Part of the problem was that the default context specified the 127.0.0.1 entry in ~/.kube/config, so I added an explicit cluster_url.

Have tested this with both ec2 and local/linux deployments.

[jmrodri]

ACK