Add JLArrays extension and tests

[kshyatt]

JLArrays.jl is a mocking package from GPUArrays.jl that allows us to test array semantics and behavior without requiring a GPU or working CUDA/ROCm install. I added an extension here so that packages which depend on StridedViews can use JLArrays to test GPU integration on machines which don't have GPUs.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
ext/StridedViewsJLArraysExt.jl	0.00%	8 Missing ⚠️

Files with missing lines	Coverage Δ
ext/StridedViewsJLArraysExt.jl	0.00% <0.00%> (ø)

🚀 New features to boost your workflow:

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

[lkdvos]

Looks good, however one main question here just to check that I am following:

Is the goal to try and rewrite/add to the Strided kernels an implementation that does not rely on scalar indexing? While I agree that it would be nice to have such functionality for the GPUarrays, I would guess that it would look like custom kernels (in the same spirit of https://github.com/JuliaGPU/GPUArrays.jl/blob/88dcf9c24b4d68e5e467135e1c6420fea459d1ae/src/host/linalg.jl#L712-L756). Does that then work with JLArrays, or should we then manually overload that to allow the scalar indexing in the mapreduce_kernel!?

[kshyatt]

Yes, the idea would be we would eventually write some custom kernels with KernelAbstractions. These can be directly used with JLArrays in a "CPU only" way, so we would be able to test that everything at least ran and gave correct results.

[kshyatt]

This is exactly what JLArrays itself does to test the GPUArrays internal kernels, etc.

[kshyatt]

Added a small test just to make sure the adapt stuff works in some sense