* reorg prestaging so that it happens on a per aggregate basis instead
  of as a phase before we start trying to create the slivers on each
  aggregate. This is mostly because of aggregates we cannot reach, and
  are deferred until we can reach them.

* Clean up of bookkeeping and handling of scheduled/deferred aggregates.

* Add simple template for defining nodes on routes, that allows setting
  the disk image and install/execute services on a per route basis. This
  will likely change to something more sophisticated. But this is good
  for the very near term.

* Bus daemon that is reading the UofU site and pinging the boss/control
  node for each bus. Update the apt_mobile_buses table with current
  location, route, and alive/dead status.

* Route daemon that is reading the apt_mobile_buses table and moving
  buses into and out of experiments as they move onto and off routes.
  Very simple for now.