and clarify some related implementation details.
Most notably, this change clarifies that the verification audit workers
and reverification audit workers are meant to live in a process or
processes separate from the satellite core, and outlines an extra queue
that will be used for communication with the core.
It's not entirely clear to me that this is the right approach; we would
save some fairly significant implementation time by leaving both types
of worker in the core. That would make it necessary to reconfigure and
restart the core when we wanted to change the number of verification
and/or reverification workers, and scaling would be limited to the
computational capacity of the core vm, but maybe those are acceptable
conditions.
Another option would be to leave the Verifier workers in the core, and
having a separate process for Reverifiers. That would be sort of a
middle way between the two above.
Change-Id: Ida12e423b94ef6088733b13d5cc58bdb78f2e93f
