[no squash] Networking improvements / hardening

[sfan5]

key points:

• resistance to flood and/or resource exhaustion (partial)
• don't be a good target for UDP amplification
• resistance to blind injection attacks and/or spoofed source IP

To do

This PR is Ready for Review.

It's best to review the commits individually.

How to test

I verified all the specialized stuff by hand. What this needs is real-world testing to catch oversights.

[SmallJoker]

I gave this PR some testing by joining several public servers. No issues found. It would be more interesting to see how well it works on a busy. Any server owner volunteers?

[lhofhansl]

FWIW. I tried locally last weekend and did not observe any problems. This is of course a limited sample.
Maybe next weekend I can try with injecting some latency into the Linux networking stack.

[LoneWolfHT]

This has been running on CTF for a week+ now (using the commits before cb5ac4d).
It works, and there are no issues as far as I can tell.

Will this cause/get put in a patch release when merged?

[sfan5]

Will this cause/get put in a patch release when merged?

Unlikely.

[SmallJoker]

The code looks clean and is reliable based on the server testing results. 👍

[Bastrabun]

Will this cause/get put in a patch release when merged?

Given that this was already used in the wild: Yes, please. Unless you're really going to fast track 5.9.0

Also, if my server can help test stuff, please ping me.