[Onnx] Switched from using pytest.skip to tvm.testing.known_failing_t…

…argets

After merging of the `tvm.testing.parametrize_targets` and
`tvm.testing._auto_parametrize_target` code paths,
`known_failing_targets` can be used in both cases.

tests/python/frontend/onnx/test_forward.py

@@ -1567,11 +1567,9 @@ def verify_hardsigmoid(input_dim, alpha, beta):
 
 
 # TODO (mbrookhart, electriclilies) Fix argmin on GPU and enable this test
+@tvm.testing.known_failing_targets("cuda")
 @tvm.testing.parametrize_targets
 def test_forward_arg_min_max(target, dev):
-    if "cuda" in target:
-        pytest.skip("Fails on CUDA")
-
     def verify_argreduce(input_dim, op_name, axis=None, keepdims=None):
         a_np1 = np.random.uniform(-10, 10, input_dim).astype(np.int32)
         out_shape = list(a_np1.shape)
@@ -5039,11 +5037,9 @@ def verify_reverse_sequence(x, sequence_lens, batch_axis, time_axis):
     verify_reverse_sequence(x, sequence_lens, 1, 0)
 
 
+@tvm.testing.known_failing_targets("cuda")
 @tvm.testing.parametrize_targets
 def test_qlinearconv(target, dev):
-    if "cuda" in target:
-        pytest.skip("Fails on CUDA")
-
     def verify_qlinearconv(
         x_shape,
         w_shape,
@@ -5341,11 +5337,9 @@ def get_random_uniform(shape, dtype="float32", high=1.0, low=0.0, seed=None):
     tvm.testing.assert_allclose(real, expected, rtol=1e-5)
 
 
+@tvm.testing.known_failing_targets("cuda")
 @tvm.testing.parametrize_targets
 def test_convinteger(target, dev):
-    if "cuda" in target:
-        pytest.skip("Fails on CUDA")
-
     def verify_convinteger(
         x_shape,
         w_shape,