2.x: Upgrade checkout to v4, setup-java to v4.1.0

[barchetta]

Description

Upgrades GitHub actions:

• setup-java to v4.1.0
• checkout to v4

See #8435

Documentation

No impact

[senivam]

Looks like newer JDK (Temurin) fails the build:
Previous:

proposed in the PR:

[senivam]

For my local build of the PR everything works just fine:

Apache Maven 3.9.5 (57804ffe001d7215b5e7bcb531cf83df38f93546)
Maven home: /home/mavines/software/maven/apache-maven-3.9.5
Java version: 11.0.22, vendor: Oracle Corporation, runtime: /media/storage/software/java/jdk-11.0.22
Default locale: en_US, platform encoding: UTF-8
OS name: "linux", version: "6.5.0-21-generic", arch: "amd64", family: "unix"

[barchetta]

The failing test is known to be unstable: #4323

[senivam]

The failing test is known to be unstable: #4323

why not to ignore it then? :D As I would say - no test no problem.

The strange thing about it is that it was failing 3 times in a row on the same JDK and all the time with different outputs.

[senivam]

To prove the issue is JDK related, I've done a tiny modification to your PR which is rolling back just one JDK version for the Helidon build:

Basically, the version of the action is not important, it does not relate to JDK's version, but they've just updated the Temurine JDK to 11.0.22. And I have to roll it back to the 11.0.21. After I did this the build just had passed OK:

Before the change it was failing in my local branch as well.

The reference to the passing action is https://github.com/senivam/helidon/actions/runs/8146385385
The reference to the code change is senivam@5ee64e9

[barchetta]

The test is unstable due to a race condition so the changing of the JDK is likely altering timing. Looks like @danielkec fixed the race condition in 3.x but not 2.x. I've back ported the fix in #8453. Hopefully that resolves the problem!