Coroutines finalizer

In case of the app crashing, we can lose coroutines which haven't been completed.
We might need to write more asynchronous code but we can't allow modules like _tags storage_ to lose new data.

We should implement global exception handler ("last frontier"), which knows about spawned coroutines and ensures they complete before crashing. The same handler would send crash reports (#78).

Start by tracing how coroutines are spawned and how the tags storage module records new data; no specific files or tests are named in the issue. Review issue #78 for the planned crash-reporting behavior. Done means a global exception handler can account for spawned coroutines, let required work complete before the app crashes, and support crash reporting.

Written by the indexing model from the issue text.