Add InterlockedOr tests

[bob80905]

This PR adds tests for the InterlockedOr HLSL function.
It tests all existing overloads of this function, including the declarations of this function as members of resources.
It also includes a fix to texture element type processing, as there was not a case to account for uint32 element types.
Fixes #101

[Icohedron]

Interesting change. Do we want to add REQUIRES: SM_6_X to every test?

Every test has a # RUN: %dxc_target -HV 202x -T cs_6_6 -Fo %t.o %t/source.hlsl line specifying the target shader model (cs_6_6 implies SM_6_6).

[bob80905]

It's definitely up for conversation. The reason I added it is because there was a machine I ran against that did not have shader model 6.6 support. So this test shouldn't run on machines that don't have GPUs that support high enough shader models.
Here is the failure: https://github.com/llvm/offload-test-suite/actions/runs/25931189782/job/76225188838#step:13:338
Its directly complaining about shader model target incompatibility, so I think the proper resolution is to add these availability features.

[farzonl]

Interesting change. Do we want to add REQUIRES: SM_6_X to every test?

Every test has a # RUN: %dxc_target -HV 202x -T cs_6_6 -Fo %t.o %t/source.hlsl line specifying the target shader model (cs_6_6 implies SM_6_6).

Hmm I hear why you did this. I'm wondering if there is a way we can unify the run line and the requires. It would suck if these fell out of sync. Maybe a cleanup task for later?

[bob80905]

Made a follow up issue here: #1219