Description of problem:
When there is  /usr/bin/openshift in containerized env, the upgrade_nodes.yaml fail at task "Verify masters are already upgraded" and report message "Master running 3.3.1.5 must be upgraded to 3.4.0.30 before node upgrade can be run." 
The root cause is that openshift_fact get current version wrongly using "/usr/bin/openshift version".  It should use "/usr/local/bin/openshift version" which is updated by openshift_container_binary_sync.py


Version-Release number of selected component (if applicable):
atomic-openshift-utils-3.4.31-1.git.0.4b78e57.el7.noarch

How reproducible:
always

Steps to Reproduce:
1. containerized install ocp 3.3 on aws.
2. install atomic-openshift-3.3.1.5 on first master which installed as a dependency of installing atomic-openshift-tests.
3. upgrade masters to v3.4
# ansible-playbook $upgrade_control_plane.yml
4. upgrade nodes  to v3.4
# ansible-playbook $upgrade_nodes.yml

Actual results:
Upgrade failed due to the current version come from /usr/bin/openshift version.

Expected results:
Upgrade node successfully.

Additional info:
upgrade log in attachment