optimize channel availability thread log output

[funky-eyes]

• I have searched the issues of this repository and believe that this is not a duplicate.

Ⅰ. Issue Description

建议改为debug下输出堆栈，info下不输出堆栈，并且将所有失败的地址合并输出，而不是10个里面有5个地址不可用输出五次日志得形式

Ⅱ. Describe what happened

If there is an exception, please attach the exception trace:

Just paste your stack trace here!

Ⅲ. Describe what you expected to happen

Ⅳ. How to reproduce it (as minimally and precisely as possible)

1. xxx
2. xxx
3. xxx

Ⅴ. Anything else we need to know?

Ⅵ. Environment:

• JDK version :
• Seata version:
• OS :
• Others:

[slievrly]

Why would an exception print is what I focus on, rather than eliminating the exception message instead.

---

为什么会触发异常的打印是我所关注的，而不是消灭异常信息。

[funky-eyes]

Why would an exception print is what I focus on, rather than eliminating the exception message instead.

为什么会触发异常的打印是我所关注的，而不是消灭异常信息。

raft模式下是很有可能的,成员中有节点下线,但是由于有自检不影响client使用,但是这个10s自检会一直输出日志,因为它在集群成员中可能没有被摘除,或者短期内无法恢复.
并且一些file作为注册中心类型,以及一些与注册中心网络通但是跟client网络存在限制的情况.我们可以允许有异常信息提示,这是必要的,但是不能老是打印出堆栈,像kafka等中间件一般是日志提示无法连接或断开连接,不会把堆栈打印

[shenshichao163-oss]

I would like to work on this issue. I have submitted PR #8203 to optimize the channel availability log output:

1. Changed LOGGER.error() to LOGGER.info() at info level (summary without stack traces)
2. Changed LOGGER.error() to LOGGER.debug() at debug level (per-address detailed stack traces)
3. All failed addresses are merged into one log entry
4. Changed else if to independent if blocks so both levels work independently

Please review when you have time. Thanks!