Test support for alternative conv layout

[pmaybank]

Overview

• currently bool cwhn = false in conv_2d test cases

• this PR adds bool cwhn = true cases

• to resolve the test failures introduced by extending the test suite,

  • the ggml_backend_cuda_device_supports_op function now returns unsupported if input or kernel layout is not contiguous.
  • the ggml_cuda_op_conv2d function tests that input layout is contiguous (previously only kernel layout was tested)

• as a result of these changes the number of CONV_2D tests goes up from 1,570 to 2,002

• ./bin/test-backend-ops test -o CONV_2D -j 1 passes with the changes.

Additional information

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: This code was not generated autonomously by AI.

[ggml-gh-bot]

Hi @pmaybank, thanks for your contribution!

Per our contribution guidelines, the automated PR checker found the following issue(s) that need your attention:

• PR Template not respected: Please respect the template when creating a new pull request. Make sure to fill out all required sections.

---

Please note that maintainers reserve the right to make final decisions on PRs. If you believe there is a mistake, please comment below.

[pmaybank]

Hi @pmaybank, thanks for your contribution!

Per our contribution guidelines, the automated PR checker found the following issue(s) that need your attention:

• PR Template not respected: Please respect the template when creating a new pull request. Make sure to fill out all required sections.

Please note that maintainers reserve the right to make final decisions on PRs. If you believe there is a mistake, please comment below.

I modified the Requirements section / AI Disclosure. Let me know if any other specific checks are failing re. the PR Template.

[jeffbolznv]

I looked at this briefly to see if it would run on vulkan. It asserts, which is fixable, but it seems like a lot of the tests are also not supported in the CPU backend, and this makes it hard to verify the other backends. Is there a plan for this?