COMP: Extend Pixi-Cxx post-build cleanup to MSVC artifacts (followup #6322)

[hjmjohnson]

Follow-up to #6322 (merged): widens the Pixi-Cxx workflow's post-build cleanup to also delete MSVC *.obj/*.lib artifacts and runs ccache --evict-older-than 1d before --cleanup so the windows-2022 runner gets the same disk relief that #6322 gave dashboard-script clients via itk_common.cmake.

Why a separate workflow change is needed

PR #6322 added the equivalent *.o/*.a cleanup to itk_common.cmake between ctest_build and ctest_test. That helps Azure DevOps / CircleCI / custom nightly clients that drive CTest via ctest -S dashboard.cmake.

The ITK.Pixi workflow runs cmake + ctest directly (no dashboard script — pixi run configure-ci then cmake --build build then ctest -j3 --test-dir build), so it does not pick up the itk_common.cmake cleanup hook. It has its own inline "Free disk space after build" step that already handled POSIX object files. This PR widens that step to also cover Windows MSVC artifacts and adds the older-than-1d ccache eviction that #6322 introduced for dashboard clients.

Diff

- find build -type f -name "*.o" -delete
- find build -type f -name "*.a" -delete
+ find build -type f -regex '.*\.\(o\|a\|obj\|lib\)$' -delete
  # Trim ccache to stay within CCACHE_MAXSIZE and remove orphaned entries
+ ccache --evict-older-than 1d 2>/dev/null ||true
  ccache --cleanup 2>/dev/null || true

[hjmjohnson]

ITK.Linux failures fixed in #6325 (dashboard files deleting mesh .obj files).
1467 - itkMeshFileReadWriteOBJTest (Failed) ITKIOMeshOBJ
1468 - itkMeshFileReadWriteOBJWithPointDataTest (Failed) ITKIOMeshOBJ
1469 - itkOBJMeshIOTest1 (Failed) ITKIOMeshOBJ
1470 - itkOBJMeshIOTest2 (Failed) ITKIOMeshOBJ

[hjmjohnson]

@dzenanz

These failures are fixed with other PR #6325

ITK.Arm64 / ARMBUILD-Python
ITK.Arm64 / ARMBUILD-Ubuntu-24.04-arm
ITK.Arm64 / ARMBUILD-x86_64-rosetta
ITK.Linux

[dzenanz]

PR 6235 is about ingestion, you must have meant some other PR, maybe #6325?

[hjmjohnson]

PR 6235 is about ingestion, you must have meant some other PR, maybe #6325?

@dzenanz Yes. Dyslexia got the best of me. #6325 not 6235.

[hjmjohnson]

windows build failure is a flake on downloading external data unrelated to this PR.

Root cause
The job is not failing because of compiler/test logic. It fails during build-time data fetching for:

GenericLabelInterpolatorRotateLabelsTest
test definition: [Modules/Filtering/GenericLabelInterpolator/test/CMakeLists.txt](https://github.com/InsightSoftwareConsortium/ITK/blob/0643ff5a34668420bf2e9879ad67db37582edeff/Modules/Filtering/GenericLabelInterpolator/test/CMakeLists.txt)
Relevant lines:

line 15: DATA{Input/classification_s4.mha}