Replace run_forever implementations with a generic one in _internal

[shsms]

This could be another reason to remove the auto-restart logic out of the actor, so we have a consistent policy for auto-restart. For example, the actor now has some sort of backoff to avoid a flood of restarts if it is "permanently" broken IIRC.

Since you are using this snippet in several places in this PR, maybe it is worth extracting the forever loop to a separate function in _internal and use it in the actor run and in the new places here? But it can be done in a separate PR too.

Originally posted by @llucax in #900 (comment)

Related issues

The solution to this issue needs to have in mind the following related issues:

• Most tasks swallow exceptions #826
• Improve the BackgroundService and Actor interface to handling internal tasks #819
• Make actor automatic restarting more resilient #632