RSS / Rack Initialize: Fail fast and explicitly

When we send the rack-initialize call, before RSS sends out the signal for bootstrap agents to initialize the rack, there should be a preliminary check done first. In this check, RSS will contact each of the sleds it expects to be in the cluster on the bootstrap addresses provided in the rack initialize call. The RSS agent will ask each of the sled agents to verify that it's ready to initialize the rack. The individual sled/bootstrap agents then check all the preconditions for initializing the sled and report back to RSS. If there are failures, all of them should be reported back to the caller so they can take remedial action. This includes not being able to contact expected sleds. This should be a fast operation, on the order of a few seconds.

cc @andrewjstone