Push Activation State Machine exception handling needs improvement

[QuintinWillison]

There are places where we catch Exception and then re-throw as uncaught RuntimeException (e.g. here), which would almost always be inappropriate and heavy handed.

Answers to this question perhaps illustrate the point better than I could describe myself, in respect as to why catch (Exception) is almost always a bad idea.

Plus, throwing a RuntimeException will cause the application process to crash. Thus, we effectively have engineered in a critical failure state when I suspect something lighter weight could have been implemented.

A customer has been seeing crashes on Android devices at runtime which pass through this exception handling path. Our manner of handling these exceptions is making it more difficult to debug that issue (internal thread).

┆Issue is synchronized with this Jira Bug by Unito