Add VLESS seed configurations

[yuhan6665]

#3260

[yuhan6665]

因为 inbound 取消了 buffer 的原因 原来的 PingPong 测试（即 inbound 收到包后立即发出现存数据或者假包）需要重写 f0e6bdf
但是这个不是必须的 所以先发一个 rebase 的分支

[0x3mp7y]

@yuhan6665 Is something actually can be done about that in seed?

net4people/bbs#546 (I understand that current observation is not clear and this needs further investigation but it's nearly impossible because it requires remote access.)

Maybe it's possible to add some sort of a queue and maxConnections? This could also be useful in general. Not sure if it's realistically possible.

Current methods could be theoretically detected as TLS-in-TLS as far as I understand.

I also remember there was some talks about Switch vision that could possibly solve such things.

[RPRX]

@0x3mp7y 我看了那个 issue，我感觉是那个运营商针对了 Vision 现有代码中的 padding 参数，你可以自行修改它们并编译、测试

不会改的话也没关系，Seed 月底就出了，并且我在考虑下个版本要不要同时修改一下默认 padding 参数，它们已经三年了

[0x3mp7y]

@0x3mp7yI've looked at that issue, and I think it's related to the carrier targeting the padding parameters in Vision's existing code. You can modify them yourself, compile, and test.

It's okay if you don't know how to change it; Seed will be released at the end of the month. I'm also considering whether to modify the default padding parameters in the next version, as they've been in place for three years.

I tried disabling flow what so ever on tests, but without adding a MUX that wouldn't help. So not sure about that.

The bad thing that it's currently temporal tests on small providers, and the guy I was able to connect to don't have such policy applied anymore.

Anyway the good thing that currently no one really checks for TLS-in-TLS :)

[RPRX]

I tried disabling flow what so ever on tests, but without adding a MUX that wouldn't help. So not sure about that.

Mux 不仅能减少连接数，还有混淆流量特征的效果，再加上你在原帖中的描述，所以我是那么猜的

The bad thing that it's currently temporal tests on small providers, and the guy I was able to connect to don't have such policy applied anymore.

小范围试点很正常，要测一下误杀概率，可能误杀率较高所以要继续优化下再重新测试

Anyway the good thing that currently no one really checks for TLS-in-TLS :)

可以试试 Trojan，反正在中国曾经是一天封一个端口，现在不知道