Fix cgroup memory enable and dashboard configuration URL

[hectcastro]

These are set of fixes that were required for me to reproduce your results on a Raspberry Pi/Kubernetes cluster of 4 nodes:

• Update mechanism to enable cgroup memory
• Fix dashboard configuration URL

Thanks for taking the time to codify these steps. Really helped speed up the process of getting a cluster up-and-running.

[chris-short]

Can you highlight the differences between cgroup_enable=memory vs. cgroup_memory=1? I have seen both in docs and am curious why it worked for me and not you. This could be very helpful in the future.

[hectcastro]

I believe that the changes are related to raspberrypi/linux@ba742b5. I started with a 4.9 kernel and kubeadm init would not initialize until I made the cgroup_memory=1 change and rebooted the Raspberry Pi nodes.

(Rebased to get rid of the merge conflict.)