[CI] Retry flaky fp8 cutlass mla tests (#24536)

Signed-off-by: Nick Hill <[email protected]>

Author: njhill

tests/kernels/test_cutlass_mla_decode.py

@@ -49,7 +49,13 @@ def cal_diff(x: torch.Tensor,
 @pytest.mark.parametrize("block_size", [64])
 @pytest.mark.parametrize("causal", [True])
 @pytest.mark.parametrize("varlen", [False, True])
-@pytest.mark.parametrize("torch_dtype", [torch.bfloat16, torch.float8_e4m3fn])
+@pytest.mark.parametrize(
+    "torch_dtype",
+    [
+        torch.bfloat16,
+        # fp8 can have occasional precision-related failures.
+        pytest.param(torch.float8_e4m3fn, marks=pytest.mark.flaky(reruns=2))
+    ])
 @torch.inference_mode()
 def test_cutlass_mla_decode(b, s_q, mean_sk, h_q, h_kv, d, dv, block_size,
                             causal, varlen, torch_dtype):