Replace use_default_config in new test cases.

Author: choijon5

test/test_errors.py

@@ -296,7 +296,7 @@ def torch_nonzero_in_device_code(x: torch.Tensor) -> torch.Tensor:
     def test_torch_chunk_device_error(self):
         """Test that torch.chunk raises error in device loops and suggests hl.split()."""
 
-        @helion.kernel(use_default_config=True, static_shapes=True)
+        @helion.kernel(autotune_effort="none", static_shapes=True)
         def kernel_with_chunk(q: torch.Tensor) -> torch.Tensor:
             _, _, M, D = q.shape
             D = hl.specialize(D)
@@ -330,7 +330,7 @@ def kernel_with_chunk(q: torch.Tensor) -> torch.Tensor:
     def test_torch_unbind_device_error(self):
         """Test that torch.unbind raises error in device loops and suggests hl.split()."""
 
-        @helion.kernel(use_default_config=True, static_shapes=True)
+        @helion.kernel(autotune_effort="none", static_shapes=True)
         def kernel_with_unbind(q: torch.Tensor) -> torch.Tensor:
             _, _, M, D = q.shape
             D = hl.specialize(D)

test/test_matmul.py

@@ -203,7 +203,7 @@ def test_matmul_packed_int4_block_size_constexpr(self):
         torch.manual_seed(0)
         M = N = K = 32
 
-        @helion.kernel(use_default_config=True, static_shapes=True)
+        @helion.kernel(autotune_effort="none", static_shapes=True)
         def matmul_bf16_packed_int4(
             A: torch.Tensor, B_packed: torch.Tensor, C: torch.Tensor
         ) -> torch.Tensor: