client端获取RPC调用实际访问过的所有的node IP,包括back request访问的node

**Is your feature request related to a problem? (你需要的功能是否与某个问题有关?)**
controller.remote_side()只能返回一个远程node的ip信息,即开启backup request的话就是最先返回的那个节点.

**Describe the solution you'd like (描述你期望的解决方法)**
希望controller提供个接口类似remote_side(),能够在client端获取经过的链路信息,这样就能够知道哪个结点高延迟的.

**Describe alternatives you've considered (描述你想到的折衷方案)**

**Additional context/screenshots (更多上下文/截图)**

Start with the controller.remote_side() API and the backup request path described in the issue. Trace how the client currently records the responding node, then define how an interface could expose every node visited during the RPC chain. Done means clients can retrieve all accessed node IPs and identify which node incurred high latency.

Written by the indexing model from the issue text.