Support random test port

[gastaldi]

By supporting random ports during test execution, this PR should allow running HTTP tests in parallel (as long as they run in different JVMs)

Fixes #2269 and closes #2324

[gastaldi]

@gsmet it would be nice if this PR could make it to 0.23.0. I'd appreciate if you (or someone else) could review

[geoand]

This is going to conflict with #4128 (which I have already had to rebase to fix a different conflict 🤪)

[gsmet]

Is there a specific reason? It looks like a new feature to me so I would keep it for 0.24.0.

[gastaldi]

@gsmet no strong reason except early feedback from users, I'm fine with that in 0.24.0 ;)

[gastaldi]

Rebased, thanks for the warning @geoand 👍

[jglick]

as long as they run in different JVMs

FYI: #3938

[jmartisk]

I just tried running the MP TCKs with it and noticed that setting a -Dquarkus.http.test-port breaks a lot of tests in MP OpenApi tests. I'll have a look what can be done about that (it's not caused by this PR).
Perhaps an additional CI check with a random port enabled would be a good thing to introduce.
Anyway very nice, this will be very useful

[geoand]

I like this! Let's see what other think

[stuartwdouglas]

LGTM, the only caveat is that this will not work with native mode tests.

The only way I can see to make it work is to parse the log output to read the http port, but that kinda sucks.