Protection from accidental slashing condition violations

[mkalinin]

Issue

Restart of validator instance potentially leads to double voting or double block production. In its turn this may be a cause of slashing condition violation.

Honest Validator doc advises to store produced blocks and attestations causes on disk in order to prevent double voting. But disk interaction adds yet another point of failure.

Possible solution

Prevent validator votes until the end of current slot. This approach, also, adds a new point of failure cause there could be a time shift between restarts. Hence, it needs to be analyzed with more details.