Send parsed watchdog host metrics to engineers pre-rotation

Your watchdog timeout events arrive unparsed, causing alert storms and untracked hosts. Give engineers standardized host metrics and timestamped incident points so issues are triaged before the next on-call rotation.