VM host that has crashed should be automatically rebooted

Definition of done: When host goes down, it should automatically reboot and not require any manual steps

Currently VM host crash is detected only by this process:

1. VM host crashes
2. After 5h developer sees "Timeout" in Coin
3. Developer does restage and is happy
4. Developer 2 sees "Timeout" in Coin
5. Developer 2 does restage and is happy
6. Developer 3 sees "Timeout" in Coin
7. Developer 3 asks in IRC about the timeouts
8. Someone from CI restarts the host

What should happen is:

1. VM host crashes
2. Coin detects that the host has crashed and restarts work items that were running on it (QTQAINFRA-1749)
3. Automatic monitoring detects the host is down and restarts the host (QTQAINFRA-1754)
4. Root cause of the problem is diagnosed/categorized and reported to CI operators (QTQAINFRA-1778)

 Definition of done for this ticket: Automatic monitoring detects the host is down and restarts the host