Allow parameterizing the nuxeo deployment strategy

This would allow for example while testing a nuxeo upgrade to use the "Recreate" strategy instead of the default one "RollingUpdate", and sequence the stop of the previous nuxeo instance before the start of the next.
Also, this is closer to a real world scenario where containers are replaced by the cloud operators.
This will ensure when the deployment is undeployed, that the previous pod is totally gone and it will avoid trying to get its logs and fail.