Remove BeaconState proto message

[syjn99]

What type of PR is this?

Other: (Very nice) cleanup

What does this PR do? Why is it needed?

This PR removes beacon_state.proto and automatically generated *.pb.go files corresponds to it. Instead, we now replaces all proto message definition of BeaconState{Fork} with native Go struct.

Before:

message BeaconState {
  // Versioning [1001-2000]
  uint64 genesis_time = 1001;
  // ... and so on
}

• Add new BeaconState per fork in beacon_state.proto file
• Run make gen for executing protoc, so that we have a Go struct.
• This system enforces us to have few converter functions from/to proto, as well as slice type even though it is completely fine to use array type for fixed bytevector.

After:

// in beacon_state_mainnet.go
// BeaconState is the Phase0 SSZ state.
type BeaconState struct {
	GenesisTime                 uint64
	GenesisValidatorsRoot       []byte `ssz-size:"32"`
	Slot                        primitives.Slot
	Fork                        *Fork
	LatestBlockHeader           *BeaconBlockHeader
// and so on

• Add new BeaconState per fork in beacon_state_{mainnet|minimal}.go file
• No need to generate *.pb.go files related to the state. Plus we remove over 9K lines of generated files which is awesome.
• Note that we don't need few entries in ssz_proto_library.bzl as it is now natively maintained in native Go struct definition.

Which issue(s) does this PR fix?

Fixes #17503

Other notes for review

How to review this PR

This PR is designed to review commit by commit. It consists of three parts:

Revisit the testing harness

Until fix(ssz): derive query field names without json tags, these are collections of groundwork for the later work. First commit (test(state): support non-protobuf states in SSZ harness) is note-worthy - as it touches our equality check. Main purpose is dropping proto dependency on equality check.

Mechanical change from Altair to Gloas, and then Phase0

From refactor(state): replace Altair protobuf with SSZ struct to refactor(state): replace Phase0 protobuf with SSZ struct,
each commit mechanically migrates to native struct per fork. Most of the diff of each commit is regenerated *.pb.go files and some equality check in our spectest runner.

Phase0 is deliberately done after Gloas as it has different wire format and pattern when saving in the storage.

Cleanups

From refactor(state): drop protobuf from beacon-chain/state, commits are mostly cleaning things up which are now unused or dead. refactor(proto): remove beacon_state.proto is note-worthy, as it retires beacon_state.proto and moves each proto message into separate parts in the codebase, so please read the description.

Production Risk

Before #17523, there was a very tiny production risk on DB migration, but now we don't have. I believe this PR doesn't bring any production risk and should work well without any problems, but here's some notes:

• DB compatability: We save BeaconState in SSZ format, so it doesn't need to be considered. Same behavior.
• CopySlice(nil) returns nil instead of [].
• SSZ Query package changes are not considered as production IMO.

Follow-ups

• Move state-related definition and function into dedicated package. proto/ package is not proper location now.
• Rename *Proto* methods. I deliberately omitted it in this PR because it covers almost entire codebase.
• Delegate SSZ List limit and SSZ Vector length to fieldparams configuration which is considered as canonical.
• Better memory allocation and no more Proto conversion.
  • As methodical-ssz doesn't support yet for array type, we need a patch there.

Acknowledgements

• I have read CONTRIBUTING.md.
• I have included a uniquely named changelog fragment file.
• I have added a description with sufficient context for reviewers to understand this PR.
• I have tested that my changes work as expected and I added a testing plan to the PR description (if applicable).