Exception in thread "Thread-18" java.lang.IllegalArgumentException: count < 0

- [ ] I have searched the [issues](https://github.com/codingapi/tx-lcn/issues) of this repository and believe that this is not a duplicate.

### 1. Bug Description
详细错误信息:
Exception in thread "Thread-18" java.lang.IllegalArgumentException: count < 0
at java.util.concurrent.CountDownLatch.(CountDownLatch.java:199)
at com.codingapi.txlcn.tc.txmsg.TMSearcher.search(TMSearcher.java:80)
at com.codingapi.txlcn.tc.txmsg.AutoTMClusterEngine.prepareToResearchTMCluster(AutoTMClusterEngine.java:83)
at com.codingapi.txlcn.tc.txmsg.AutoTMClusterEngine.onInitialized(AutoTMClusterEngine.java:57)
at com.codingapi.txlcn.tc.txmsg.TCSideRpcInitCallBack.lambda$null$1(TCSideRpcInitCallBack.java:95)
at java.util.ArrayList.forEach(ArrayList.java:1257)
at com.codingapi.txlcn.tc.txmsg.TCSideRpcInitCallBack.lambda$connected$2(TCSideRpcInitCallBack.java:95)
at java.lang.Thread.run(Thread.java:748)

为什么启动集群配置，启动正常成功，但是会出现这样的异常信息？

### 2. Environment:
- JDK version:1.8
- OS:window10或centos7.8
- TX-LCN version:5.0.2
- Others:

### 3. Exception Stacktrace

```
Paste your Exception Stacktrace here!
```

### 4. Tour Idea

Start at com.codingapi.txlcn.tc.txmsg.TMSearcher.search, especially TMSearcher.java:80, and trace the count passed during AutoTMClusterEngine.prepareToResearchTMCluster and onInitialized. Reproduce cluster startup on JDK 8 with TX-LCN 5.0.2 on Windows or CentOS, then identify why the count becomes negative and verify startup completes without the exception.

Written by the indexing model from the issue text.