Optimize Msg Structures

[GalRogozinski]

Description

Investigate how we can create better message structures.
Here are some ideas:

• Move preconsensus justifications out of consensus data (or delete them entirely)
• In PartialSignatureMessages we can move signing roots to header ssv-spec#600
• remove individual PartialSignatureMessage signer field ssv-spec#135
• Remove duty and version out of consensus data
• RunnerRole can be compressed farther on wire ssv-spec#423
• Refactor QBFT Identifier as type with fixed length ssv-spec#536

[diegomrsantos]

Suggestion: use fixed-size encoding for RSA signatures

SignedSSVMessage.Signatures is currently defined as:

Signatures  [][]byte  `ssz-max:"13,256"`

Since RSA-2048 signatures are always exactly 256 bytes (output is always the size of the modulus), the inner dimension could use fixed-size encoding instead of variable-length. The current ssz-max:"13,256" allows signatures shorter than 256 bytes at the wire level, which are never valid for RSA-2048.

Using fixed-size inner encoding would:

• Enforce the invariant at the SSZ layer
• Produce slightly more compact encoding (no offset pointers per signature)
• Eliminate the need for runtime signature size validation

This came up during review of structural validation in Anchor — the VariableList<u8, U256> type (matching the Go spec) only guarantees <= 256, not == 256, so exact-size checks must be done at validation time rather than being enforced by the type system.