seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒

### Check Ahead

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

- [ ] I am willing to try to implement this feature myself.

### Why you need it?

按照seata官网部署了seata raft高可用集群，并发量为500TPS，发现GlobalBeginRequest、GlobalCommitRequest这两个阶段耗时很长，长达100-500多毫秒，seata-server给到的8C/8G 硬盘为ssd，
max-append-bufferSize: 524288 ，比官网增加1倍
disruptor-buffer-size: 32768 比官网大1倍
不知还可以调整哪些哪些参数，降低请求处理时长

### How it could be?

_No response_

### Other related information

_No response_

Start with the GlobalBeginRequest and GlobalCommitRequest handling paths and the reported Seata Raft settings, then reproduce the issue at 500 TPS while measuring each stage. Done means identifying the latency cause and validating a configuration adjustment or fix against the reported 100–500 ms delays.

Written by the indexing model from the issue text.