Migrate from capture_pre_autograd_graph to torch.export.export_for_training (#5730)

Summary:
Pull Request resolved: #5730

As titled. The `capture_pre_autograd_graph` API is deprecated.

Reviewed By: hsharma35

Differential Revision: D63541800

fbshipit-source-id: 7b830ae55a5dff8bf61be0470f54302c2ab461d8
(cherry picked from commit b4a6148)

Author: mcremon-meta

backends/cadence/aot/compiler.py

@@ -30,7 +30,6 @@
 )
 from executorch.backends.transforms.remove_clone_ops import RemoveCloneOpsTransform
 from executorch.exir import EdgeCompileConfig, EdgeProgramManager, to_edge
-from torch._export import capture_pre_autograd_graph
 from torch.ao.quantization.pt2e.export_utils import model_is_exported
 from torch.ao.quantization.quantize_pt2e import convert_pt2e, prepare_pt2e
 
@@ -58,7 +57,7 @@ def convert_pt2(
     """
 
     # Export with dynamo
-    model_gm = capture_pre_autograd_graph(model, inputs)
+    model_gm = torch.export.export_for_training(model, inputs).module()
 
     if model_gm_has_SDPA(model_gm):  # pyre-fixme[6]
         # Decompose SDPA