Standardize solo machine signature format in proto definitions

[colin-axner]

Summary

As mentioned during the pr review for solo machine, the signature format should use a standardized proto definition instead of concatenating byte string.

Proposal

Create proto definitions for each type of signature needed to verify counterparty state.

Some questions I still have:

• do headers need timestamps?
• do we want to allow extra data to be included in the signature (an additional []byte field to each definition)?

message SignBytes {
  uint64 sequence = 1;
  bytes data = 2; // proto marshaled definitions below
}

message HeaderData {
  google.protobuf.Any new_pub_key = 2;
}

message ClientStateData {
  uint64 timestamp = 1;
  bytes path = 2;
  google.protobuf.Any client_state = 3; 
}

message ConsensusStateData {
  uint64 timestamp = 1;
  bytes path = 2;
  google.protobuf.Any consensus_state = 3; 
}

message ConnectionStateData {
  uint64 timestamp = 1;
  bytes path = 2;
  ibc.connection.ConnectionEnd connection = 3; 
}

message ChannelStateData {
  uint64 timestamp = 1;
  bytes path = 2;
  ibc.channel.Channel  channel = 3; 
}

message PacketCommitmentData {
  uint64 timestamp = 1;
  bytes path = 2;
  bytes commitment = 3; 
}

message PacketAcknowledgementData {
  uint64 timestamp = 1;
  bytes path = 2;
  bytes acknowledgement = 3; 
}

message PacketAcknowledgementAbsenseData {
  uint64 timestamp = 1;
  bytes path = 2;
}

message NextSequenceRecvData {
  uint64 timestamp = 1;
  bytes path = 2;
  uint64 next_seq_recv = 3;
}

Because we need to know the sequence signed to prove misbehaviour, I added the SignBytes layer to encapsulate the non-sequence signed data.

---

For Admin Use

• Not duplicate issue
• Appropriate labels applied
• Appropriate contributors tagged
• Contributor assigned/self-assigned

[cwgoes]

do headers need timestamps?

They can have timestamps, at minimum we should not reset the timestamp to zero on a header update.

A client's timestamp should never decrease.

do we want to allow extra data to be included in the signature (an additional []byte field to each definition)?

I don't think this is necessary but cc @warner

[fedekunze]

I can work on this if you want @colin-axner

[colin-axner]

@fedekunze go for it 👍

[cwgoes]

Also, perhaps at the same time, we should add the signature diversifier per cosmos/ibc#476.