Make is_sparse a property of MaskedTensor (pytorch#110725)

Fixes pytorch#104574

Seeing that MaskedTensor is a prototype, the BC breaking nature of this change seems okay?

Locally tested:
<img width="1372" alt="image" src="https://github.com/pytorch/pytorch/assets/31798555/239e61ba-e0b9-4909-8c7a-0ce3869d7375">

Pull Request resolved: pytorch#110725
Approved by: https://github.com/cpuhrsch

torch/masked/maskedtensor/core.py

@@ -331,5 +331,6 @@ def is_sparse_csr(self):
         return self.layout == torch.sparse_csr
 
     # Update later to support more sparse layouts
+    @property
     def is_sparse(self):
         return self.is_sparse_coo() or self.is_sparse_csr()

torch/masked/maskedtensor/reductions.py

@@ -46,7 +46,7 @@ def _torch_reduce_all(fn):
     def reduce_all(self):
         masked_fn = _get_masked_fn(fn)
         data = self.get_data()
-        mask = self.get_mask().values() if self.is_sparse() else self.get_mask()
+        mask = self.get_mask().values() if self.is_sparse else self.get_mask()
         # When reduction is "all", then torch.argmin/torch.argmax needs to return the index of the
         # element corresponding to the min/max, but this operation isn't supported correctly for sparse layouts.
         # Therefore, this implementation calculates it using the strides.
@@ -67,7 +67,7 @@ def reduce_all(self):
             result_data = torch.sum(idx * stride)
 
         # we simply pass in the values for sparse COO/CSR tensors
-        elif self.is_sparse():
+        elif self.is_sparse:
             result_data = masked_fn(masked_tensor(data.values(), mask))
 
         else:
@@ -80,7 +80,7 @@ def reduce_all(self):
 
 def _torch_reduce_dim(fn):
     def reduce_dim(self, dim, keepdim=False, dtype=None):
-        if self.is_sparse():
+        if self.is_sparse:
             msg = (
                 f"The sparse version of {fn} is not implemented in reductions.\n"
                 "If you would like this operator to be supported, please file an issue for a feature request at "