Each observation message should create a heartbeat

[KaraMelih]

Especially when testing, we do not send periodic heartbeats. So what happens is, we trigger an alert but the server thinks that there are no available detectors, and thus the false-alarm computation function struggles (well, it crashes; see SNEWS2/SNEWS_Coincidence_System#100)

We should be attaching a heartbeat each time we receive a coincidence message.

[habig]

One side effect I can see happening from this is that when a signal goes in, the heartbeat rate might jump up and cross the notification threshold.

Which isn't a problem, the contact will get an email, but that contact is either testing and knows it, or would appreciate an extra email from SNEWS since his experiment is probably screaming "SUPERNOVA" and knowing the signal went in is a nice side effect.

[KaraMelih]

hmm. I think the heartbeat warnings are sent only if the heartbeat is not received within mu+3sigma but I don't think anything happens if there is an early heartbeat within mu-3sigma.

[KaraMelih]

Being implemented in this PR #116

[habig]

Ahh ok! So no worries then.