Description of problem:
ovirt-etl currently doesn't stop on the case of connection refuse or any other error, it just logs the error and tries again. In order to prevent ovirt-etl from stopping and starting only by a user intervention, we need to implement a WatchDog.

How reproducible:
always

Steps to Reproduce:
1. Casue a error to ovirt-etl (shutdown potgres for instance).
2. Check the ovirt-etl status & tail /var/log/ovirt/ovirt-dwh/ovirt-etl.log 

Actual results:
ovirt-etl is not stopped, only error is logged.

Expected results:
ovirt-etl should die on error and start automatically when the error state is resolved (tested with the watchdog). Watchdog should log it as started ovirt-etl from error state.