Gemm/Fma kernel on avx512 use new solution, with less permute and shuff instruction

StgSharp.Native/sn-head/sn_intrinsic.context.x86.h

@@ -1,3 +1,6 @@
+#ifndef SN_INTRINSIC_CONTEXT_X86_H
+#define SN_INTRINSIC_CONTEXT_X86_H
+
 #include "sn_target.h"
 
 #if SN_IS_ARCH(SN_ARCH_X86_64)
@@ -271,3 +274,4 @@ INTERNAL int SN_DECL index_pair_sse(short const *stram, uint32_t target, int len
 typedef enum intrinsic_level { SIMD0 = 0 } intrinsic_level;
 
 #endif
+#endif