This disables AVX2 usage on x86_64 (by defining NO_AVX_HARDWARE) as well as
removes now-unused includes for AVX2/AVX512 instrinsics (the NEON header
was never used in the first place and the way it's included is completely
wrong anyway).

We need it disabled on x86_64 because we're generic (and AVX is sandybridge
and newer, but the enabled code was for haswell and newer), and on other
arches it's disabled for obvious reasons.

--- Clp/src/ClpPackedMatrix.cpp
+++ Clp/src/ClpPackedMatrix.cpp
@@ -6749,11 +6749,6 @@ ClpPackedMatrix3::ClpPackedMatrix3()
 }
 #ifdef _MSC_VER
 #include <intrin.h>
-#elif defined(__arm__)
-#include <arm_neon.h>
-#else
-#include <immintrin.h>
-//#include <fmaintrin.h>
 #endif
 /* Constructor from copy. */
 ClpPackedMatrix3::ClpPackedMatrix3(ClpSimplex *model, const CoinPackedMatrix *columnCopy)
@@ -6777,7 +6772,7 @@ ClpPackedMatrix3::ClpPackedMatrix3(ClpSimplex *model, const CoinPackedMatrix *co
 {
   //#undef COIN_AVX2
   //#define COIN_AVX2 8
-  //#define NO_AVX_HARDWARE
+#define NO_AVX_HARDWARE
 #ifndef COIN_AVX2
 #define COIN_AVX2 4
 #else
--- Clp/src/ClpSimplexDual.cpp
+++ Clp/src/ClpSimplexDual.cpp
@@ -3556,11 +3556,6 @@ void moveAndZero(clpTempInfo *info, int type, void *extra)
 #endif
 #ifdef _MSC_VER
 #include <intrin.h>
-#elif defined(__arm__)
-#include <arm_neon.h>
-#else
-#include <immintrin.h>
-//#include <fmaintrin.h>
 #endif
 int ClpSimplexDual::dualColumn0(const CoinIndexedVector *rowArray,
   const CoinIndexedVector *columnArray,