Fix type annotation in _ConvNd for in_channels (pytorch#84302)

`_ConvNd` has an attribute `in_channels` that was mistakenly annotated as `_in_channels`.

This fixes pytorch#84223
Pull Request resolved: pytorch#84302
Approved by: https://github.com/albanD

torch/nn/modules/conv.py

@@ -54,7 +54,7 @@ class _ConvNd(Module):
     def _conv_forward(self, input: Tensor, weight: Tensor, bias: Optional[Tensor]) -> Tensor:
         ...
 
-    _in_channels: int
+    in_channels: int
     _reversed_padding_repeated_twice: List[int]
     out_channels: int
     kernel_size: Tuple[int, ...]