[Feature Request] Support Multi-Agent Routing from Single Channel Endpoint

## Summary

Support routing messages to different agents from a single channel endpoint.

## Component(s) Affected

- [x] Core / Backend (app, agents, config, providers, utils, local_models)
- [ ] Console (frontend web UI)
- [x] Channels (DingTalk, Feishu, QQ, Discord, iMessage, etc.)
- [ ] Skills
- [ ] CLI
- [ ] Documentation (website)
- [ ] Tests
- [ ] CI/CD
- [ ] Scripts / Deploy

## Problem / Motivation

Currently, one channel instance can only bind to one agent. When integrating with external platforms that have multiple bots or robots, users must deploy separate QwenPaw instances for each agent.

## Proposed Solution

Enable a single channel to distribute incoming messages to different agents based on message metadata.

## Alternatives Considered

1. Deploy multiple QwenPaw instances
2. Use chat_with_agent tool for forwarding
3. Use external reverse proxy for routing

## Additional Context

Current behavior: All messages from a channel are processed by the same agent.
Expected behavior: Messages can be routed to different agents based on their properties.

## Willing to Contribute

- [x] I am willing to open a PR for this feature (after discussion).

Start by tracing the Core / Backend channel handling and how incoming message metadata is exposed to agent selection; the issue names no files or tests, so the relevant entry points must first be located. Done means one channel can route messages to different agents by their properties without requiring separate QwenPaw instances, with coverage for the routing behavior.

Vom Indexierungsmodell aus dem Issue-Text verfasst.