单个节点设置backup request timeout导致超时不符合预期

**Describe the bug**
如果我通过name service初始化一个channel（比如bns），这个name service只解析出来一个节点，我设置这个channel的超时是10ms，backup request 超时是2ms。当我使用这个channel发起rpc调用2ms触发backup request超时的时候channel尝试选取另一个节点失败，会导致rpc在2ms左右的时候失败，这个感觉不符合预期，我设置的是10ms超时才对。

**To Reproduce**

**Expected behavior**

**Versions**
OS:
Compiler:
brpc:
protobuf:

**Additional context/screenshots**

No source file or test is named. Start by tracing channel timeout handling and backup-request node selection for a name-service channel that resolves to one node, with a 10ms channel timeout and 2ms backup-request timeout. Done means the RPC does not fail at about 2ms solely because no alternate node can be selected, and instead respects the configured overall timeout.

Written by the indexing model from the issue text.