[bug] Stale connection retries

[St4NNi]

Bug

If the internal notifier connection was not used for approx. 1 minute it goes stale and needs to be retried on the next occurrence, this results in an "Internal Notifier Error" error.

To fix this we should implement 2 measures:

1. Add retry / backoff logic for random transport errors in tonic / h2
2. Open a new backend connection for each user request

[St4NNi]

This will be mitigated by: arunaengine/aruna@6ffb29a

For a proper retry logic we should wait for: hyperium/tonic#733
and: tower-rs/tower#682