If the docker daemon goes down and comes up, it will try to restart
containers. This will race with our own restart loop. Only one loop should
be controlling restarts, probably.
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.
