Establish convention for capturing system incidents

[grahamalama]

We are trying to reduce the frequency of incidents which cause customers to ask, "why isn't my bug syncing?". In order answer the question "are we reducing the frequency?", we need to capture this data consistently.

In an ADR, we should establish a workflow for tracking incidents, including the start time and resolution time. This way, we can capture both the number of incidents and the average time to resolve incidents.

A non-exhaustive list of "incidents" might include:

• webhook queue becomes disabled
• a workflow or workflows record n partial syncs over some duration of time
• a workflow or workflows record n total sync failures over some duration of time