Consensus v2.3

[nazar-pc]

This tries to collect implementation steps necessary to get to consensus v2.3. The idea is to outline major individual steps that need to be resolved (in many cases this can happen in parallel). More detailed separate issues can be created from these individual steps and can even have lower level steps behind them. Consider this issue as an epic.

Implementation steps (non-exhaustive)

Cryptographic primitives

Preview Give feedback

1. Polynomial erasure coding (based on FFT, like v2.2 farmer plotting) subspace/subspace#1163
   consensus core farmer +3
   
2. polynomial coefficients recovery subspace/subspace#1307
   consensus improvement upstream +3
3. Chia proof-of-space table generation and API around it for querying table and verification as described in the specification subspace/subspace#1164
   consensus core farmer +3
   
4. Replace KZG implementation with the same primitives used for polynomial erasure coding subspace/subspace#1280
   breaking-consensus consensus core improvement node +5
   

Archiver updates

Preview Give feedback

1. Erasure coding we use right now should be replaced with polynomial-based subspace/subspace#1282
   breaking-consensus consensus core node +4
   
2. Archiver API needs to become incremental subspace/subspace#1160
   node performance +2
   

Farmer updates

Preview Give feedback

1. Implement farming/plotting/verification/retrieval according to v2.3 consensus subspace/subspace#1349
   breaking-consensus breaking-runtime consensus core farmer node +6
   

Runtime

Preview Give feedback

1. Host functions exposing KZG crypto in a performant way (especially because we'll increase max degee and size of public parameters with it) subspace/subspace#1279
   breaking-runtime consensus core enhancement node +5
   
2. Probably host functions for erasure coding for the same reason would need to be exposed too
3. Probably host functions for Chia proof-of-space for the same reason would need to be exposed too

More tasks to come.

[dariolina]

FYI, I have removed task "Polynomial erasure coding where we can specify evaluation points" because we decided to go with a more classical approach and start evaluation points from 0 as usual. At least for the prototype, but hopefully forever. If that changes, I will create a new task for it.