[Test] Skip ROCm AITER MLA prefill tests on non-ROCm platforms (#49945)

Signed-off-by: Liangliang-Ma <liangliang.ma@intel.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

tests/v1/attention/test_mla_prefill_selector.py

@@ -8,6 +8,7 @@ import pytest
 import torch
 
 from vllm.config import AttentionConfig, ModelConfig, VllmConfig
+from vllm.platforms import current_platform
 from vllm.platforms.interface import DeviceCapability
 from vllm.v1.attention.backends.mla.prefill.base import MLADimensions
 from vllm.v1.attention.backends.mla.prefill.registry import MLAPrefillBackendEnum
@@ -287,6 +288,11 @@ class TestBackendValidation:
             assert invalid_reasons == []
 
 
+@pytest.mark.skipif(
+    not current_platform.is_cuda_alike(),
+    reason="Imports vllm.platforms.rocm, whose module init requires a CUDA or "
+    "ROCm torch build; not importable on XPU/CPU/TPU.",
+)
 class TestROCmAiterFAPrefillSelection:
     """Tests for the ROCm AITER FlashAttention MLA prefill backend."""