Rescheduling continuous audits from FAILED nodes

This patch set adds background job that reschedules CONTINUOUS
audits from FAILED to ACTIVE decision engine nodes using round-robin
algorithm. It also contains fix for main[1] HA PS about filtering audits.

[1]: https://review.openstack.org/#/c/578102/

Partially-Implements: blueprint support-watcher-ha-active-active-mode
Change-Id: Ib248a6cd3adbd3927c47db6bb819300361492411

watcher/tests/decision_engine/test_scheduling.py

@@ -62,7 +62,7 @@ class TestCancelOngoingAudits(db_base.DbTestCase):
             goal=self.goal,
             hostname='hostname1',
             state=objects.audit.State.ONGOING)
-        cfg.CONF.set_override('host', 'hostname1')
+        cfg.CONF.set_override("host", "hostname1")
 
     @mock.patch.object(objects.audit.Audit, 'save')
     @mock.patch.object(objects.audit.Audit, 'list')