Execute simple JUnit tests and @QuarkusComponentTest first

[gsmet]

I was hoping to constrain JUnit tests and @QuarkusComponentTests but @QuarkusComponentTest is in another artifact so this will have to do.

Fixes #46035.

Note that you can still break things if you specify another priority but that's how it is :).

[geoand]

Rebased on main since the errors were weird and I want to see if they were actually caused by the PR

[gsmet]

Yeah, I need to check what's going on. Will see if I can get it before we branch 3.19 but I'm not sure about that.

[gsmet]

@geoand so there's something leaking somewhere but the test infra of the module was all borked. I isolated the ITs in a proper Failsafe run instead of grouping them with the Surefire tests and things are fine now.

The problem was caused by the leaking tests now being executed first as they are ITs and not having any sort of Quarkus test annotation.

At some point, we should probably try to figure out what the leak is but given Holly's ongoing work, I'm not sure it's worth it.

[geoand]

At some point, we should probably try to figure out what the leak is but given Holly's ongoing work, I'm not sure it's worth it.

Right. At best it could be redundant, at worst it could cause more merge conflicts

[quarkus-bot]

Status for workflow Quarkus CI

This is the status report for running Quarkus CI on commit 1abb1b9.

✅ The latest workflow run for the pull request has completed successfully.

It should be safe to merge provided you have a look at the other checks in the summary.

You can consult the Develocity build scans.

[holly-cummins]

I've had a quick peek at this and #46035. I think:

• In my new code, the 'last' app isn't closed, same as now. So I don't think QuarkusTest TestProfile pollutes environment of QuarkusComponentTest #46035 would have worked with my branch before @gsmet's fix
• I think there's a chance that my new code would give better isolation to the ITs that Guillaume had to isolate, but I haven't actually checked
• I have a suspicion this change is responsible for an ecosystem CI failure in the Pact extension, [CI] - Pact + Quarkus main quarkiverse/quarkiverse#94 (comment); there was already some cross talk between tests and super-fragile name-based ordering in the Pact suite, and this probably perturbed that. Weirdly, the symptom is a classloading error, and I have no idea why test cross-talk can cause a classloading error, but I've been successfully ignoring the problem for some time. :)