- General performances improvement (25%) on insert/update/collideTT.

- ~30% additional performance improvement for win32 using SSE implementation for critical methods.

src/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp

@@ -112,7 +112,6 @@ void	Process(const btDbvtNode* na,const btDbvtNode* nb)
 //
 btDbvtBroadphase::btDbvtBroadphase(btOverlappingPairCache* paircache)
 {
-btDbvt::benchmark();
 m_releasepaircache	=	(paircache!=0)?false:true;
 m_predictedframes	=	2;
 m_stageCurrent		=	0;