Fix node storage corruption automatically

[pasqu4le]

Clarification and motivation

octez-node can sometime exit with leaving a corrupted state on disk, which happens more frequently if it's interrupted abruptly.

The node can sometimes fix the state by itself, but it can also fail to do so, preventing it from starting.

This affects also nodes launched with our systemd services, which can fail to start for this reason.

It should be possible to automatically detect the failure of the node to fix its corrupted state and consequently perform a cleanup of the node storage and a re-sync of it.

Acceptance criteria

When our tezos-node service is started with a corrupted state that octez-node cannot fix by itself, then the service fixes it instead.