heptapod_runner.paas_dispatcher: singled out a `decommission()` method

Decommission should not even be done synchronously in the event
processing thread, and this will help doing it in another thread.

Meanwhile, it helps keeping `process_events` small for inspection.

branch : heptapod
1 job for topic/heptapod/state-persistence in 1 minute and 39 seconds (queued for 2 seconds)
Status Name Job ID Coverage
passed test-python #251588