Place the service file (or a link to it) in /etc/systemd/system/ Place the watchdogged.py file somewhere ( and change the ExecStart portion in the .service to point at the file )
then do systemctl daemon-reload
followed by systemctl start watchdogged.service
After this you can watch the progress using journalctl --follow -u watchdogged.service
change the PROBABILITY variable to something else to watch it faster/later or succeed.