Fixes #343 withincode doesn't work as specified

[jafarre-bi]

Pointcuts of the form withincode(MethodSignature) don't match method-execution joinpoints, which is in contradiction with the documentation. Moreover, both constructor-execution and initialization joinpoints are matched by pointcuts of the form withincode(ConstructorSignature). This agrees with the documentation, but is inconsistent with the behaviour regarding method execution.
This PR allows withincode pointcuts to match method-execution joinpoints and modifies the corresponding unit tests.

[jafarre-bi]

My doubt here is whether withincode should match initialization joinpoints or not, as explained in the comment. What do you think @kriegaex?

[kriegaex]

@jafarre-bi, Being inactive at the moment, I really have not looked into your issue and this PR, validating whether there is in fact a real problem that needs to be solved, maybe just a documentation issue, or neither. But if you still think that your PR solves a problem, please note that it causes test failures. Please check if either your code or the tests need to be adjusted.

[jafarre-bi]

@jafarre-bi, Being inactive at the moment, I really have not looked into your issue and this PR, validating whether there is in fact a real problem that needs to be solved, maybe just a documentation issue, or neither. But if you still think that your PR solves a problem, please note that it causes test failures. Please check if either your code or the tests need to be adjusted.

Yes, that's why it's still a draft. I didn't fully understand how that worked. Now I do.
In the end, it's simply adding method execution to the joinpoint kinds matched by withincode pointcuts.
Surprisingly, there are unit tests checking specifically that they don't match. Difficult to understand, as it's inconsistent with constructor and initialization pointcuts, which do match, and directly in contradiction with the documentation.
I just pushed the changes with unit-test fixes. The tests don't pass in my local environment because of unexpected warnings from the compiler, which are probably due to the version of my JDK. Let's see if they pass here.

[kriegaex]

Let's see if they pass here.

They did. Next, you probably want to take a look at the failing integration tests.

[kriegaex]

Regarding 26caa9f, there are still failing ITs. Furthermore, you seem to have committed a few files with only whitespace changes. You should fix up the commit to no longer contain those changes and force-push. BTW, I am still not saying anything about the changes in the code and tests, because I really have not looked into any of it.

[jafarre-bi]

In the first commit, I committed indentation that mixed spaces with the original tab-based indentation, and it wasn't aligned. That's why there are whitespace-only changes in the second commit. If you look into the changes as a whole, there are no whitespace-only changes.

[jafarre-bi]

I finally managed to get all tests to pass in my local environment by changing the JDK to version 21. I also removed all changes in white space.
While I believe this is a bug fix, it may also be a breaking change, as some pointcuts will match joinpoints they didn't match before. However, the more I think about it, the clearer it becomes that the problem isn't with the documentation. It would not be easy to document the current behaviour solidly.
There's something else I haven't looked into yet. I took a quick look at the implementation of the @withincode annotation's behaviour, and I believe it currently behaves differently from the withincode clause. It appears to exclude all container-type joinpoints, even constructor execution and initialisation. The difference would be more pronounced with the changes in this PR.