refactor: Simplify peer identity types (PeerId cleanup)

[sanity]

Background

Part of #2164 (peer identity restructuring). The behavioral fixes for NAT traversal are complete in PRs #2167 and #2169, but the type system still has some redundancy.

Current State

We have overlapping types for identifying peers:

Type	Contains	Purpose
PeerId	{addr: SocketAddr, pub_key: TransportPublicKey}	Legacy bundle
PeerKeyLocation	{pub_key, peer_addr: PeerAddr, location}	Combined info + location
SocketAddr	Network address	Routing
TransportPublicKey	Cryptographic key	Identity

The Issue

PeerId bundles address and identity together, but:

• Identity (pub_key) is immutable and unique
• Address (SocketAddr) can change (NAT, roaming)

This conflation was the root cause of NAT issues fixed in #2164. The behavioral fix is done (connection-based routing), but the type still exists and is used in ~256 places.

Proposed Cleanup

Option A: Make PeerId a type alias for TransportPublicKey

• pub type PeerId = TransportPublicKey;
• Use PeerKeyLocation when you need both identity and address
• ~137 compile errors to fix

Option B: Remove PeerId entirely

• Replace with TransportPublicKey for identity
• Replace with SocketAddr for address
• Replace with PeerKeyLocation for combined
• Similar effort to Option A

Scope

• ~256 usages across 20+ files
• Main files: p2p_protoc.rs, connection_manager.rs, handshake.rs
• Estimated: 2-3 hours of mechanical refactoring

Priority

Low - the behavioral fixes are complete and tested. This is purely type system hygiene.

Related

• Depends on: refactor(routing): add upstream_addr for connection-based routing #2167, refactor(connect): remove from/sender fields from ConnectMsg #2169 (should be merged first)
• Part of: refactor: Restructure peer identity and address handling #2164

[AI-assisted - Claude]

[github-actions]

🤖 Auto-labeled

Applied labels:

• T-enhancement (95% confidence)
• E-easy (90% confidence)
• A-networking (90% confidence)
• A-developer-xp (75% confidence)

Reasoning: This is a codebase cleanup/refactor to simplify peer identity types (no user-facing bug or new feature), so T-enhancement fits. The author estimates ~2-3 hours of mechanical refactoring, so E-easy is appropriate. The change concerns peer identity/address types used in networking and connection/handshake code, so A-networking is primary; it also improves developer experience and type hygiene, so A-developer-xp is relevant.

Previous labels: none

If these labels are incorrect, please update them. This helps improve the auto-labeling system.