* Improve the initial nrvam setup
* More fallbacks
* Rebuild the initial setup and upgrade of a node.
This is mostly about removing duplicated effort and data, which should
make it easier to add things in the future.
* Re-setup and repair the node on each reboot.
By running this every time we reboot, we have a chance to repair problems in the configuration
automatically. We can detect changes and force a reboot only when necessary.
* Improve startup and upgrade
* Remove debug delays
* Move after telnet (for emergency access)
* Ignore ubootenv file when diffing
Tim Wilkinson