By default we run retirement_check for Vms, Services, LoadBalancers and
OrchestrationStacks every 10 minutes.

Most of the time, retirement_check results into no-op. However, it has
fixed costs attached.
 - it inserts, selects updates, and deletes from MiqQueue
 - it locks miq_queue_lock
 - it synces log messages (they grow)
 - it takes worker's time to process

Every 10 minutes.

Moreover, the code base is very similar -> no point of having it so
many times. Every method defined increases our footprint. Every penny
counts.

Even in cases, your set-up uses retirement dates heavily, each run will
go through really quickly -> so there is not reason for separation. The
work is already too small, to divide into multiple jobs.

See: https://www.pivotaltracker.com/story/show/145237979

Define the scope in the RetirementMixin

Create a scope for items not scheduled for retirement