Add per-LID peer routing for Honcho memory

Problem

In multi-user WhatsApp deployments, one gateway account can receive messages from multiple senders. The Honcho memory plugin currently resolves identity from a default peer or gateway runtime identity, which is not enough when a deployment needs to map sender LIDs to stable Honcho peer names.

Fix

Add optional peerByLid support:

1. plugins/memory/honcho/client.py: add a peer_by_lid config field loaded from honcho.json.
2. plugins/memory/honcho/session.py: resolve a known LID from the session key before falling back to existing peer resolution.
3. plugins/memory/honcho/__init__.py: when user_id contains a known LID, rewrite it to the configured peer name before session manager initialization.

Example configuration:

{
  "hosts": {
    "your-host": {
      "peerByLid": {
        "1234567890": "alice",
        "9876543210": "bob"
      }
    }
  }
}

The change is opt-in. Without peerByLid, the existing fallback behavior is unchanged.

Files touched

• plugins/memory/honcho/client.py
• plugins/memory/honcho/session.py
• plugins/memory/honcho/__init__.py

Verification

• Syntax checked with python3 -m py_compile on the three touched files.
• Suggested follow-up: add unit tests for two configured LIDs and the empty-config fallback.