Make masked_scatter core aten (pytorch#137949)

Summary: Making `masked_scatter` core aten since it is hard to decompose and we now have a portable kernel for it

Test Plan: N/A

Differential Revision: D64368725

Pull Request resolved: pytorch#137949
Approved by: https://github.com/larryliu0820

Author: jackzhxng

aten/src/ATen/native/native_functions.yaml

@@ -8015,6 +8015,7 @@
   variants: function, method
   dispatch:
     CompositeExplicitAutograd: masked_scatter
+  tags: core
 
 - func: masked_scatter_backward(Tensor grad_output, Tensor mask, SymInt[] sizes) -> Tensor
   dispatch: