Bug: entire committee slot fails due to slashing protection instead of affected validator

[olegshmuelov]

#1902 introduced a fix preventing attesting at the epoch it was registered or reactivated
and not by slashing protection as it was before causing the whole committee slot to fail.

In rare case below we still can bypass the minParticipationEpoch protection

1. If a validator (e.g., Validator X) is registered at the very first slot of an epoch.
2. Immediately after the registration, there is a restart (e.g., due to a node crash or manual restart).
3. Validator X has a duty at slot n (later in the same epoch) as part of a committee with other validators.
   In this scenario, the restart would potentially result in the failure of all committee slots because the state needed to recognize Validator X’s registration might not persist correctly across the restart. This could lead to missed duties for the entire committee

It might be worth considering additional safeguards to handle state persistence during restarts or improving recovery logic to avoid this failure scenario.

Originally posted by @olegshmuelov in #1898 (comment)