Mid-term protection for multistaking

[AurelienFT]

Context

Currently we are lacking of protection against multi-staking. We need to have some protections for the mainnet before implementing the solution that our research team developed : https://arxiv.org/pdf/2302.10009.pdf

Current implementation

Whenever a valid block (or a header) is received we will add it to our consensus, it will be processed and the cliques calculations will happens. In case of thousands of blocks at the same slot we could store and process thousands of cliques

Mid term solution

We will implement some rules that will prevent multiples blocks for the same slot to be processed

Current implementation

We receive a block or a header through massa-protocol and we send it to massa-consensus and the consensus make the clique calculation with this block. (If it's just a header massa-consensus ask to massa-protocol to retrieve the rest of the block) If we miss some parents of this block massa-consensus ask to massa-protocol to retrieve them

Changes implemented

The first two header (or full block) received for a slot are saved and follow the current workflow. Having two blocks for a slot that are propagated allow all the network to be aware that there is a double stake.

Any block after this two received for the same slot will not be processed by consensus.

Point of attention

If we receive a block that has as a parent one the block we refuse at previous slot because it was a third one we need to still be able to ask it through the consensus and not be blocked by protocol that have saved him in cache etc...

[AurelienFT]

@damip ping for review

[AurelienFT]

Done in #3646