Optimize the dirty dependency graph walk

[aloubyansky]

The current implementation may appear to be hanging processing (very) dirty dependency graphs. This change checks whether a dependency node has been already processed before walking its dependencies.

Fix #27589

[knutwannheden]

Using the Collections.newSetFromMap() utility is sometimes also nice :-)

final Set<ArtifactKey> visited = Collections.newSetFromMap(new IdentityHashMap<>());

[quarkus-bot]

Failing Jobs - Building 067e888

Status	Name	Step	Failures	Logs	Raw logs
✔️	JVM Tests - JDK 11
✖	JVM Tests - JDK 11 Windows	Build	Failures	Logs	Raw logs
✔️	JVM Tests - JDK 17
✔️	JVM Tests - JDK 18

Full information is available in the Build summary check run.

Failures

⚙️ JVM Tests - JDK 11 Windows #

- Failing: extensions/opentelemetry/opentelemetry/deployment 
! Skipped: extensions/opentelemetry/opentelemetry-exporter-jaeger/deployment extensions/opentelemetry/opentelemetry-exporter-otlp/deployment integration-tests/micrometer-prometheus and 6 more

📦 extensions/opentelemetry/opentelemetry/deployment

✖ io.quarkus.opentelemetry.deployment.RestClientOpenTelemetryTest.urlWithoutAuthentication line 111 - More details - Source on GitHub

org.opentest4j.AssertionFailedError: expected: <CLIENT> but was: <SERVER>
	at org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
	at org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)

[gsmet]

@aloubyansky I think the failure is unrelated. @radcortez it might be related to the same issue with CONSUMER timing we had elsewhere?

@aloubyansky I don't know if you want to change the Set/Map thing so I let you merge when you want.

[radcortez]

@radcortez it might be related to the same issue with CONSUMER timing we had elsewhere?

Maybe. I did some digging here #27368.

And the puzzling part is that when the test fails, some spans have the same startTime, which messes with the expected order. That startTime is set by the OTel library, so we don't control that. I had a look and I didn't see anything out of the ordinary. I'm wondering if the server updates the SO time in the middle of the test and then it messes the spans, but it does seem a little far-fetched.