FIX: Make global queue objects thread-local

[david-cortes-intel]

Description

Sycl queues on GPU are managed through global objects, but since the introduction of array API, some operations might be offloaded to other libraries that raise the GIL during their operations, and might thus lead to conflicts if sklearnex objects are used in parallel through python threads instead of being in separate processes.

This PR fixes potential thread racing issues by making the queues thread-local, with a helper class as suggested by @ahuber21 .

---

Checklist:

Completeness and readability

• I have commented my code, particularly in hard-to-understand areas.
• Git commit message contains an appropriate signed-off-by string (see CONTRIBUTING.md for details).
• I have resolved any merge conflicts that might occur with the base branch.

Testing

• I have run it locally and tested the changes extensively.
• All CI jobs are green or I have provided justification why they aren't.

[david-cortes-intel]

/intelci: run

[david-cortes-intel]

/intelci: run

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
onedal/utils/_sycl_queue_manager.py	82.35%	3 Missing ⚠️

Flag	Coverage Δ
azure	80.56% <83.33%> (+0.01%)	⬆️
github	82.24% <83.33%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
sklearnex/_device_offload.py	84.61% <100.00%> (ø)
onedal/utils/_sycl_queue_manager.py	81.81% <82.35%> (+0.99%)	⬆️

... and 4 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[david-cortes-intel]

/intelci: run

[david-cortes-intel]

/intelci: run

[david-cortes-intel]

/intelci: run

[david-cortes-intel]

/intelci: run

[david-cortes-intel]

CI error on CPU appears to happen elsewhere too, likely not related to the changes here.