[Vulkan] Test Feature/TypedBuffer/GetDimensions.test XPASSes on Windows Vulkan AMD DXC but XFAILs everywhere else

[Icohedron]

All Vulkan workflows running test Feature/TypedBuffer/GetDimensions.test XFAILs as expected, except for Windows Vulkan AMD DXC which XPASSes. This is related to #469

Should the XFAIL be changed to specifically exclude the combination of Vulkan+AMD+DXC? Or should we just leave the XFAIL as-is?

Results from scheduled runs today (2025-11-28):

╭───┬──────────────────────┬─────────────┬─────────────────────────────┬────────┬────────────────────────────────────────╮
│ # │      timestamp       │   run-id    │          workflow           │ status │                  test                  │
├───┼──────────────────────┼─────────────┼─────────────────────────────┼────────┼────────────────────────────────────────┤
│ 0 │ 2025-11-28T18:06:34Z │ 19771064332 │ Windows Vulkan AMD Clang    │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 1 │ 2025-11-28T20:04:28Z │ 19772907221 │ Windows Vulkan Intel Clang  │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 2 │ 2025-11-28T22:01:55Z │ 19774580028 │ Windows Vulkan Intel DXC    │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 3 │ 2025-11-28T22:06:03Z │ 19774646176 │ Windows Vulkan QC Clang     │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 4 │ 2025-11-28T06:02:32Z │ 19755520578 │ Windows Vulkan NVIDIA Clang │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 5 │ 2025-11-28T02:35:49Z │ 19752462306 │ Windows Vulkan NVIDIA DXC   │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 6 │ 2025-11-28T18:02:29Z │ 19770990351 │ Windows Vulkan QC DXC       │ XFAIL  │ Feature/TypedBuffer/GetDimensions.test │
│ 7 │ 2025-11-28T18:05:50Z │ 19771050617 │ Windows Vulkan AMD DXC      │ XPASS  │ Feature/TypedBuffer/GetDimensions.test │
╰───┴──────────────────────┴─────────────┴─────────────────────────────┴────────┴────────────────────────────────────────╯

[damyanp]

Need to bottom out on what is actually correct here. It looks like all other Vulkan implementations round up to 16. Maybe this is correct behavior in Vulkan, and in this case the AMD result is actually incorrect. We should confirm this!