Retry policy on polling

[BodySplash]

For now, we fail fast. For every single error, we shutdown the signer. This is good, but in the case of temporary network errors, not so much.
So the goal here would be to introduce a retry policy for polling tezos and ethereum