Reputation score for Spark checker nodes and participants

[bajtos]

Recently introduced committees and majorities (see #59) created an opportunity for building reputation scores at the level of individual checker nodes (Station ID) and also at the level of participants (wallet address = multi-node operator).

The idea is to calculate how many measurements were in the minority vs majority. "Good" nodes and participants should report measurements that are in the majority most of the time.

We need to consider how to incorporate other existing measurement evaluation criteria into the score, e.g. penalise nodes that perform too many tasks per round or tasks that were not assigned to them in this round.

[bajtos]

Related:

• Incentivise long-living Station IDs #122
• Persist state (and Station ID) across Docker upgrades node#444