Merge v0.11.x into main

[lukaszstolarczuk]

Deliveres into main:

• v0.11.2
• [CI] Use extra param to set CUDA_PATH in gpu workflow #1312
• retrieve numanode object by os index instead of logical index. #1309

[Copilot]

Pull Request Overview

This PR merges v0.11.x into main, delivering v0.11.2 and addressing issues #1312 and #1309.

• Update NUMA node queries by switching to hwloc_get_numanode_obj_by_os_index for both source and destination memory targets.
• Adjust workflow build parameters for Windows by redefining the environment variables and CMake prefix path configuration.

Reviewed Changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated no comments.

File	Description
src/memtargets/memtarget_numa.c	Switch from hwloc_get_obj_by_type to hwloc_get_numanode_obj_by_os_index for improved NUMA node retrieval.
.github/workflows/reusable_gpu.yml	Update build parameter setup by replacing CUDA_PATH with EXTRA_CMAKE_PATH and modifying CMAKE_PREFIX_PATH.

Files not reviewed (1)

• ChangeLog: Language not supported

Comments suppressed due to low confidence (3)

src/memtargets/memtarget_numa.c:269

• Since the API call has been updated to hwloc_get_numanode_obj_by_os_index, consider revising the error message to accurately reflect the new function used.

LOG_PERR("Getting HWLOC object by type failed");

src/memtargets/memtarget_numa.c:276

• Updating this error message to mention hwloc_get_numanode_obj_by_os_index will improve clarity if an error occurs during the destination NUMA node query.

LOG_PERR("Getting HWLOC object by type failed");

.github/workflows/reusable_gpu.yml:96

• Verify that using a semicolon as a delimiter for CMAKE_PREFIX_PATH is valid across all targeted platforms, as the change might affect path resolution.

-DCMAKE_PREFIX_PATH="${{env.VCPKG_PATH}};${{env.EXTRA_CMAKE_PATH}}"

[lukaszstolarczuk]

needed to add a commit with params passing fix on Windows (see the last commit)