Remove unused pass and test to replace linalg.vector_norm.

Summary:
Disable failing test for unused pass that replaced aten linear.vector_norm with a custom operator.
We no longer get this operator in edge program so the test and the pass aren't needed anymore.

```
# Input program:
def forward(x):
  return torch.linalg.vector_norm(x)

# Input size:
x = torch.randn(32)
```

```
# Edge program:
def forward(self, x: "f32[32]"):
  aten_pow_tensor_scalar: "f32[32]" = executorch_exir_dialects_edge__ops_aten_pow_Tensor_Scalar(x, 2);  x = None
  aten_sum_dim_int_list: "f32[]" = executorch_exir_dialects_edge__ops_aten_sum_dim_IntList(aten_pow_tensor_scalar, None);  aten_pow_tensor_scalar = None
  aten_pow_tensor_scalar_1: "f32[]" = executorch_exir_dialects_edge__ops_aten_pow_Tensor_Scalar(aten_sum_dim_int_list, 0.5);  aten_sum_dim_int_list = None
  return (aten_pow_tensor_scalar_1,)
```

Differential Revision: D73235672

Author: hsharma35

backends/cadence/aot/replace_ops.py

@@ -1806,30 +1806,6 @@ def call_operator(self, op, args, kwargs, meta):
         return super().call_operator(op, tuple(new_args), kwargs, meta)
 
 
-@register_cadence_pass(CadencePassAttribute(opt_level=0))
-class ReplaceAtenLinalgVectorNormWithCadenceLinalgVectorNormPass(ExportPass):
-    """
-    Replace the aten.linalg_vector_norm op with a custom op.
-    aten.linalg_vector_norm is not supported by Jarvis, so we
-    need to replace it with native_batch_norm at all optimization levels.
-    """
-
-    def call_operator(self, op, args, kwargs, meta):
-        if op != exir_ops.edge.aten.linalg_vector_norm.default:
-            return super().call_operator(op, args, kwargs, meta)
-
-        assert (
-            len(args) == 1
-        ), "aten.linalg_vector_norm should have 1 argument (a tensor), we do not support any custom variants"
-
-        return super().call_operator(
-            exir_ops.edge.cadence.linalg_vector_norm.default,
-            args,
-            kwargs,
-            meta,
-        )
-
-
 @register_cadence_pass(CadencePassAttribute(opt_level=1))
 class ReplaceSingleElementTensorArgumentsFromFullOpWithScalarPass(ExportPass):
     """
@@ -2243,7 +2219,6 @@ class CadenceReplaceOpsInGraph:
         ReplacePT2DequantWithCadenceDequantPass,
         ReplaceSingleElementTensorArgumentsFromFullOpWithScalarPass,
         ReplaceAtenAvgPoolWithJarvisAvgPoolPass,
-        ReplaceAtenLinalgVectorNormWithCadenceLinalgVectorNormPass,
         ReplaceWhereWithFullArgsWithWhereScalar,
         # ReplaceGeluWithApproximateGeluPass,
     ]

backends/cadence/aot/tests/test_replace_ops_passes.py

@@ -23,7 +23,6 @@
     MakeSliceAndCatDimOutermostPass,
     ReplaceAddMMWithLinearPass,
     ReplaceAtenConvolutionWithJarvisConvolutionPass,
-    ReplaceAtenLinalgVectorNormWithCadenceLinalgVectorNormPass,
     ReplaceConstantPadNdWithSlicePass,
     ReplaceConvolutionOptionalArgsWithConcreteArgsPass,
     ReplaceConvWithIm2RowAndLinear,
@@ -1189,36 +1188,6 @@ def forward(self, x):
             count_node(graph_after_passes, exir_ops.edge.aten.transpose_copy.int), 0
         )
 
-    def test_replace_aten_linalg_vector_norm_with_cadence_linalg_vector_norm(self):
-        class LinalgVectorNorm(torch.nn.Module):
-            def forward(self, x: torch.Tensor):
-                return torch.linalg.vector_norm(x)
-
-        x = torch.randn(32)
-
-        graph_module = (
-            export_to_edge(LinalgVectorNorm(), (x,)).exported_program().graph_module
-        )
-
-        p = ReplaceAtenLinalgVectorNormWithCadenceLinalgVectorNormPass()
-        graph_after_passes = cast(PassResult, p(graph_module)).graph_module
-
-        # Assert that aten.linalg_vector_norm op was replaced by a
-        # cadence.linalg_vector_norm op
-        self.assertEqual(
-            count_node(
-                graph_after_passes,
-                exir_ops.edge.aten.linalg_vector_norm.default,
-            ),
-            0,
-        )
-        self.assertEqual(
-            count_node(
-                graph_after_passes, exir_ops.edge.cadence.linalg_vector_norm.default
-            ),
-            1,
-        )
-
     def test_replace_aten_where_with_cadence_where_Scalar(self):
         class WhereScalarModel(torch.nn.Module):
             def forward(self, cond: torch.Tensor):