Revert "Chore: Three small 'fixes'"

[Fangliding]

revert #4922
这个commit会导致goroutine泄露~~(已确认)~~ 假的
说了没bug别乱改 乱改又改出bug还得费事修 关了好多个 终于懒得管了几次 然后这就坏了

[RPRX]

以后较新的贡献者想修 bug 要求先开可复现的 issue 吧，不能幻想出一个 bug 然后虚空修复

[patterniha]

这个commit会导致goroutine泄露(已确认)

Why do you say it's confirmed when it's not yet confirmed?‍
Racist messages in the group are shameful, Instead of solving the problem, they ...

ok, revert all my PRs and live with sniff-infinite-loop, udp-close-not-belongs, compact-send-corrupted-data, ...

I'm not saying I'm perfect, maybe the problem is with one of my PRs.
anyway, I'm working on this problem and I wrote some tests, @RPRX Please delay new-release ​for a 2-3 days until the problem is resolved.

[Fangliding]

Why do you say it's confirmed when it's not yet confirmed?‍ Racist messages in the group are shameful, Instead of solving the problem, they ...

damn, some one said revert can fix but after hours he come back and say not,so I closed this
I'm testing myself

[Fangliding]

我写了一个压测测试udp性能 最后得到了和其他人差不多的pprof结果
但是最后切断所有连接再手动触发gc后内存还是按期望回落了 感觉只是单纯客户端打开了很多连接导致的压力大 这些反馈还是有问题 但是自己复现又太麻烦太麻烦了

[RPRX]

反正我只能说，*ray 里有些代码我都不想动也不敢动也无法审核，除非我把它重构了

[patterniha]

@Fangliding test #5050