gsutil cp is prone to failure during initial creation of a Kubernetes
cluster. This 'until' loop will keep trying as suggested in the 'gsutil
cp' documentation.
Fixed up some scripts to be more robust. Changed the e2e test setup to use g1-small instances. Fixed up documentation to reflect the new script locations. Disabled the "curl | bash" cluster launch as it hasn't been well tested and doesn't include the cloudcfg tool yet.