add a unified SVM consistency check test

[bashbaug]

Adds a unified SVM consistency check test, as per the description in #2150.

Note that the target branch for this PR is the cl_khr_unified_svm branch, not the main branch.

[bashbaug]

Hmm, the CI build checks are going to fail unless we update the scripts to pull the headers from KhronosGroup/OpenCL-Headers#269.

[bashbaug]

I've updated the test as we discussed.

I also pulled in the CI changes so the unified SVM headers are used, and now the automated builds are working again.

[EwanC]

I've not been following the unified svm test plan closely, but is the idea that the extension tests will live in test_conformance/SVM rather than test_conformance/extensions/cl_khr_unified_svm?

[bashbaug]

Good question. I don't think we've discussed this. I'll create a new test executable if that's what the group prefers, but I think we've generally tried to minimize test executables, so I'm inclined to put the unified SVM tests here unless we have a good reason to do otherwise.

[bashbaug]

Hmm, I may have found a few good reasons to create a new test executable:

1. The "SVM" tests run an InitCL function to decide whether to run the tests, which currently requires OpenCL 2.0 or newer, and requires that the query for CL_DEVICE_SVM_CAPABILITIES returns a non-zero value. The first part is debatably OK, though the extension is currently written to support OpenCL 1.2 devices also. The second part is not OK though, because a device can support unified SVM without supporting SVM.

2. The "SVM" tests run with forceNoContextCreation set to true, which means no context or queue are created. Not the end of the world, but I think the harness context and queue will be sufficient for most of the unified SVM tests, so it'd be nice to keep things simple and use them.

[bashbaug]

Merging (to the cl_khr_unified_svm branch), as discussed in the February 11th memory TSG.