Crash process if failure escalates to a Guardian (esp. /system)

[ktoso]

If any of "important" system actors die, we must kill the process; this is to be handled by escalating errors all the way to the guardians who make this decision.

Same about user actors if error reached a guardian?

[ktoso]

This partially exists right now, so this is only about exposing that supervision strategy to users in API.

[ktoso]

We now have build in .escalate which acomplishes that as a first class concept <3

[ktoso]

Escalation works and is merged