Docs: add spatial shape constraints to SegResNetDS docstring (fixes #6771)

Author: Talhax55z

monai/networks/nets/segresnet_ds.py

@@ -254,6 +254,21 @@ class SegResNetDS(nn.Module):
                     image spacing into an approximately isotropic space.
                     Otherwise, by default, the kernel size and downsampling is always isotropic.
 
+                    Spatial shape constraints:
+    The input spatial dimensions must be divisible by ``2 ** (len(blocks_down) - 1)``.
+    With the default ``blocks_down=(1, 2, 2, 4)`` (4 levels), each spatial dimension
+    must be divisible by 8.
+
+    Use :py:meth:`shape_factor` to query the required divisors for a given configuration,
+    and :py:meth:`is_valid_shape` to check whether a specific input tensor satisfies them.
+
+    Example::
+
+        model = SegResNetDS(spatial_dims=3, blocks_down=(1, 2, 2, 4))
+        print(model.shape_factor())  # [8, 8, 8]
+        # Valid input: shape (1, 1, 128, 128, 128) -- all dims divisible by 8
+        # Invalid input: shape (1, 1, 100, 100, 100) -- 100 is not divisible by 8
+
     """
 
     def __init__(