Chain should alert when we stop getting data

[ssfrr]

If we stop getting sensor data from a device that was previously giving data, we should send an alert email.

The logic would be to have a "failed" flag per-device. If we haven't heard from a device in a while we should set the "failed" flag. When a device becomes failed and wasn't failed before, we should send an alert.

We need to define the timespan after a device becomes failed that triggers an alert, maybe some multiple of the sampling rate (which we'd also need to add as a sensor property).

Anyways, this needs some discussion but would be a big help in knowing when things go down, so we're not scrambling later on because we didn't realize things were down.

[kkleidal]

This could be incorporated with #29 . If we have the server writing to a stream when events are triggered, we could have a subscribed microservice which reads the stream of events and notices when events regarding certain devices stop coming (and then sends an email/SMS/tweet/etc). The microservice architecture would make the system more ready for change.