Minor simplifications and clean-ups in Toolchain processing

[HannesWell]

Replace deprecated DefaultJavaToolChain by JavaToolchainImpl.

Initially part of #2379, but there seem to be some bugs in this change.
@laeubi do you directly have any hot hint?

[github-actions]

Test Results

   561 files  ±0     561 suites  ±0   4h 27m 3s ⏱️ + 14m 4s
   363 tests ±0     357 ✔️ ±0    6 💤 ±0  0 ❌ ±0 
1 089 runs  ±0  1 070 ✔️ ±0  19 💤 ±0  0 ❌ ±0 

Results for commit fbdd082. ± Comparison against base commit 7a94e1c.

♻️ This comment has been updated with latest results.

[laeubi]

@HannesWell I have no immediate idea, especially as these are maven classes, you cna try to even split this more into some simplifications that are maybe less critical, beside that, in the past it has worked well to abstract the maven parts (e.g by doing as much as possible in tycho-core) instead of any of the mojos,, I even using reflection on some places :-\

[laeubi]

@HannesWell do you plan to work on this or should we postpone this until Maven 4?

[HannesWell]

I temporarily reverted the replacement of the deprecated DefaultJavaToolChain. Lets see if this makes a difference. I also checked again the other changes and actually they look save, but I might have overseen something.

[HannesWell]

I temporarily reverted the replacement of the deprecated DefaultJavaToolChain. Lets see if this makes a difference. I also checked again the other changes and actually they look save, but I might have overseen something.

OK, that worked. But I don't understand why it breaks something if we just use the super-class.
Lets try it again and the previous failure was just a temporary issue.
If it still fails, lets submit this with the deprecated DefaultJavaToolChain in place.

[HannesWell]

Looks like this worked now.
Could it be that the test-runners where using a too old Maven version?
Any way, @laeubi any objection to merge this now?

[laeubi]

Could it be that the test-runners where using a too old Maven version?

I have no clue, as it works now I don't see a reason to further wait with this.