Optimize out aten.detach (#378)

This op is no-op with disabled gradient.  Removing its instances should
be fine in a valid graph.

https://pytorch.org/docs/stable/generated/torch.Tensor.detach.html

Co-authored-by: Artem Yerofieiev <[email protected]>

Author: jdh8

torch_ttnn/passes/lowering/to_tt_pass.py

@@ -323,6 +323,9 @@ def __init__(self, target, args, kwargs):
             # TODO: no ttnn op can convert _adaptive_avg_pool2d
             return self.call_function_prop_meta(target, args, kwargs)
 
+        if target == torch.ops.aten.detach.default:
+            return args[0]
+
         return self.call_function_prop_meta(target, args, kwargs)