Add JMH Benchmarks for the critical code paths

[aepfli]

Our tool is excellent, but more and more people are using it. We already received internal information, that some code paths tend to be hotspots, especially structure operations.

Evaluating this without a proper test harness around it is hard; hence, it might be a good idea to add a JMH Suite to our tests for verification and to allow us to spot hotspots earlier in the future.

• https://github.com/openjdk/jmh
• https://www.baeldung.com/java-microbenchmark-harness

[toddbaert]

Yes this is a great idea. We could run these as part of the SDK CI.

We may also want to add something similar to the contribs eventually, though there it will be up to each component owner what constitutes adequate performance.

[toddbaert]

I'm not sure if #1156 fully resolves this, but it's certainly related.

@aepfli I'm happy to add a check that will check we don't have regressions with this number, but it will involve a bit of a hacky script.

[aepfli]

Do you think we can close this as #1156 is merged, or do you want to invest further in it? If yes, we should maybe adapt the ticket description to add more context and concrete to-dos.

I should start correctly reading comments. Sorry for this. Well, I am fine for now. Some verification would be good, or some persistence of at least the latest results, so we can at least verify from time to time. Furthermore, it would be cool if we found a tool where we can add this metric correctly to draw nice graphs and see how the performance evolves. But without a proper tool to display the data, I don't see a benefit in adding it to the CI.

[aepfli]

i think with #1182 and #1178 this can be closed now