Should the coordinator check inputs as far as possible?

[real-or-random]

This is related to #9, but only somewhat.

At the moment, the coordinator mostly just relays messages. The verification of the messages is left to the participants, who need to perform it anyway. The coordinator could also check messages. Would this provide any advantages?

[jonasnick]

Some notes after discussing this offline:

The advantage this would provide is that a potential bug would surface earlier. Also, the participant could replace the ParticipantOrCoordinatorError with a CoordinatorError in every case except when the received share is inconsistent with the VSS commitment.

The disadvantages are:

• Increased complexity of the coordinator.
• Duplicate checks. The participants still need to do the same checks they do right now. Duplicate checks lead to a (probably insignificant) runtime increase.

My current stance is that I am not in favor of this change because of the increased complexity and duplicate checks. If implementors have the desire to make their coordinator check inputs as far as possible, they're free to do so because this would still be consistent with the spec.