Make unrecoverable events in controller configurable

[amisevsk]

Description

As of PR #549, there is a list of events that cause the controller to mark DevWorkspaces as failed. However, some of these events may be temporary states that can be safely ignored (e.g. a pod fails to schedule for a brief period before succeeding).

We should consider making the list of unrecoverable events customizable in order to avoid unnecessary workspace failures.

Note however that whether these events should be ignored depends on how they occur (i.e. if scheduling fails for ~3-5 seconds it's fine, but if it's longer workspaces could hang for a long time, in which case it makes sense to fail the DevWorkspace).