iris: INTEL_COMPUTE_CLASS causes gpu hangs on MTL platforms
The compute queue was enabled in Intel CI, to catch regressions while mesa waits for a working GuC to be shipped with an i915 that allows mesa to query GuC versions.
On meteorlake platforms, iris will generate a gpu hang if the compute queue is enabled, running many tests including:
spec/glsl-4.30/execution/built-in-functions/cs-op-assign-mult-vec2-vec2