Full 3.2 CI run for JDK 11 and JDK 17 doesn't finish in 6 hours

[rsvoboda]

Describe the bug

Full 3.2 CI run for JDK 11 and JDK 17 doesn't finish in 6 hours limit imposed by GH actions

We have hit that on #36697 where JDK 17 run almost finished and JDK 11 run had around ~100 modules to go.

Last one for JDK 11: 2023-10-30T21:51:02.3030129Z [INFO] Building Quarkus - Integration Tests - Scala 3.2.999-SNAPSHOT [989/1088]
Last one for JDK 17: 2023-10-30T21:53:55.4269482Z [INFO] Building Quarkus - Integration Tests - Management Interface with authentication 3.2.999-SNAPSHOT [1088/1089]

CI needs to improve to be able to process bigger backports in sane way.

Expected behavior

CI runs are finished within 6 hours

Actual behavior

Full CI run for JDK 11 and JDK 17 doesn't finish in 6 hours limit imposed by GH actions

How to Reproduce?

Do something similar to #36697

Output of uname -a or ver

No response

Output of java -version

No response

Quarkus version or git rev

No response

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[rsvoboda]

We will probably need Gradle Enterprise build cache for 3.2 branch.
I don't think #34597 is the only piece we need though.

[rsvoboda]

I'm checking logs of finished and unfinished runs and turning into conclusion that the runs are primarily affected by the agent / hypervisor it runs on.

Failing run: Time elapsed: 3.693 s - in io.quarkus.resteasy.reactive.server.test.customproviders.NoTargetTest from quarkus-resteasy-reactive-deployment
But in passing run the time is 2.06 s, tests from this module do not use external services to affect the times of test execution.

Another example:

[261] Quarkus - Reactive Routes - Deployment | 00:02:17:552
   vs.
[261] Quarkus - Reactive Routes - Deployment | 00:01:46:334

(using .github/ModuleBuildDurationReport.java from this repo)