Skip Bfloat16 support when building for VSX (pytorch#61630)

Summary:
Copy-paste ifdef guard from vec256/vec256.h
Probably fixes pytorch#61575

Pull Request resolved: pytorch#61630

Reviewed By: janeyx99

Differential Revision: D29690676

Pulled By: malfet

fbshipit-source-id: f6d91eadab74bcbcb1dc9854ae1b98a0dccacd14

Author: malfet

aten/src/ATen/cpu/vec/vec256/functional.h

@@ -1,4 +1,6 @@
 #pragma once
 
 #include <ATen/cpu/vec/vec256/functional_base.h>
+#if !defined(__VSX__)  || !defined(CPU_CAPABILITY_VSX)
 #include <ATen/cpu/vec/vec256/functional_bfloat16.h>
+#endif