[transfix]

Follow-up to #774. package-windows's Run tests step pointed MAGICK_CODER_MODULE_PATH, MAGICK_FILTER_MODULE_PATH and MAGICK_CONFIGURE_PATH at the deps prefix for the whole ctest run. Since #673 and #774, libcvc finds the extracted cvcpkg ImageMagick's coders and config by itself (set_magick_paths_win in src/cvc/image/magick_io.cpp). Setting the variables only hid that lookup from every test except magick_bootstrap_test, which removes them itself.

This PR removes the block and leaves a comment explaining why the variables are absent. The whole Windows suite then runs the way a local build does, including ImageTest.* and the CvcCliTest vol2img/img2vol cases, whose cvc.exe child processes inherit the environment.

Why this should be green

• Recipe job evidence: recipe / windows-2022 / Release never set these variables (its log has no MAGICK_ at all). On image: find ImageMagick's coders beside a build tree on Windows; test with no MAGICK_* set #774 it passed all 4389 tests, including all 28 ImageTest.* and all 4 MagickBootstrapTest.* cases.
• Local evidence: on image: find ImageMagick's coders beside a build tree on Windows; test with no MAGICK_* set #774 I ran image_test, cvc_cli_test (Vol2ImgExportsSlices, Img2VolImportsSlices) and magick_bootstrap_test with no MAGICK_* set, in four Windows layouts. All passed, and ctest passed 34/34.
• No direct callers: nothing else in the tree calls Magick++ without going through cvc::init_imagemagick(). src/cvc/image and volume_ops.cpp are the only users; the mesher hits are VolMagick::.
• Dead branch: the removed block's debug\bin branch never ran, because the Windows matrix is Release only (there are no windows-debug dep bundles). $isDebug stays, since the PATH setup above still uses it.

Checked locally

• ci.yml still parses as YAML.
• The step's PowerShell, with the ${{ … }} expressions substituted, parses with 0 errors, the same as on master.