Determine how p2p event sync should work

[ryardley]

This is a placeholder issue for working out how the network should manage restart during outages and retrieving historical p2p messages (not evm)

[ryardley]

At a guess - we should probably try something along the lines of storing a certain cache of recent messages on kademlia. There are suggestions of minimising node traffic by only gossiping message hashes between nodes whilst storing the message content on kademlia. Nodes could then either store some kind of list of historical message ids in kademlia. How this list would work I am not sure. I can imagine a hash chain should guarantee some kind of integrity acknowledging there is no consensus guarantees. The event publisher could be responsible for determining the most recent event to trail. Investigating the use of prolly trees to provide a kind of auto consensus could also be interesting.

Properties:

• Low internode traffic
• Some kind of historical message order
• Avoid a single node storing the whole list