Fix gcc.target/aarch64/sve/sel_3.c for big-endian targets

A pasto in this test meant that we needed extra reverse instructions
for big-endian targets.

2020-01-22  Richard Sandiford  <richard.sandiford@arm.com>

gcc/testsuite/
	* gcc.target/aarch64/sve/sel_3.c (permute_vnx4sf): Take __SVFloat32_t
	rather than __SVFloat16_t

gcc/testsuite/ChangeLog

 2020-01-22  Richard Sandiford  <richard.sandiford@arm.com>
 
+	* gcc.target/aarch64/sve/sel_3.c (permute_vnx4sf): Take __SVFloat32_t
+	rather than __SVFloat16_t
+
+2020-01-22  Richard Sandiford  <richard.sandiford@arm.com>
+
 	* g++.dg/ext/sve-sizeless-1.C: Don't expect an error for alias
 	templates.
 

gcc/testsuite/gcc.target/aarch64/sve/sel_3.c

@@ -82,7 +82,7 @@ permute_vnx8hf (__SVFloat16_t x, __SVFloat16_t y)
 **	ret
 */
 __SVFloat32_t
-permute_vnx4sf (__SVFloat16_t x, __SVFloat16_t y)
+permute_vnx4sf (__SVFloat32_t x, __SVFloat32_t y)
 {
   return __builtin_shuffle ((vnx4sf) x, (vnx4sf) y, (vnx4si) MASK_8);
 }