Currently delete_health_monitor() emulates behaviour of
cascade deletion: when a HealthMonitor instance is deleted,
all corresponding PoolMonitorAssociations are queried and
deleted one by one. This can be done automatically by means
of SQLAlchemy if we set proper cascade flags on the relationship
between HealthMonitor and PoolMonitorAssociations models.
Fixes bug 1169107.
Change-Id: I674c381192719a56433f67be53a32203c36ebe2b
ee50162b27