refactor(logging): preserve tracebacks and add channel context

- Preserve tracebacks: logger.error in except blocks → logger.exception
- Channel context: BaseChannel injects self.logger = logger.bind(channel=name)
- Third-party bridge: redirect_lib_logging() replaces ad-hoc stdlib-to-loguru bridges
- Log levels: network timeouts downgraded from ERROR → WARNING
- Fix --verbose flag to actually work with loguru (set handler to DEBUG)

nanobot/agent/subagent.py

@@ -250,7 +250,7 @@ class SubagentManager:
         except Exception as e:
             status.phase = "error"
             status.error = str(e)
-            logger.error("Subagent [{}] failed: {}", task_id, e)
+            logger.exception("Subagent [{}] failed", task_id)
             await self._announce_result(task_id, label, task, f"Error: {e}", origin, "error", origin_message_id)
 
     async def _announce_result(