implement prioritized disputes selection

[EclesioMeloJunior]

Description

Disputes are an important inherent data component, so we should implement the algorithm that selects relevant disputes and new votes, so we can provide to the runtime relevant information.

How the prioritization works?

Generally speaking disputes can be described as:

• Active vs Inactive
• Known vs Unknown onchain
• Offchain vs Onchain
• Concluded onchain vs Unconcluded onchain

Provisioner fetches all disputes from dispute-coordinator and separates them in multiple partitions, see struct PartitionedDisputes. Each partition has got a priority implicitly assigned to it and the disputes are selected based on this priority (e.g. disputes in partition 1, then if there is space - disputes from partition 2 and so on)

Flow

1. Retrieve on chain disputes, if the runtime does not support the Disputes method then return empty
2. Retrieve recent disputes, these are disputes offchain tracked by disputes-coordinator subsytems (filter confimed, concluded and disputes that already exists on chain)
3. Partition recent disputes
4. Vote selection, selects the votes from PartitionedDisputes which should be sent to the runtime, votes which are already known on chain are filtered out, the result should be sorted by (Session Index, Candidate Hash) (runtime requirement).

Links

• https://github.com/paritytech/polkadot-sdk/blob/71c19e4c7ca2b411722ec7d9c1392e1bd81b5681/polkadot/node/core/provisioner/src/disputes/prioritized_selection/mod.rs#L89