For MSVC builds, implement 'rint()' and 'rintf()' to supplement the ones that were already implemented

msvc_extra_headers/ardourext/float_cast.h.input

@@ -56,6 +56,18 @@
 		return intgr ;
 	}
 
+	__inline double
+	rint (double flt)
+	{	long long int intgr;
+
+		_asm
+		{	fld flt
+			fistp intgr
+			} ;
+
+		return (double) intgr ;
+	}
+
 	__inline long int
 	lrintf (float flt)
 	{	int intgr;
@@ -68,6 +80,18 @@
 		return intgr ;
 	}
 
+	__inline float
+	rintf (float flt)
+	{	int intgr;
+
+		_asm
+		{	fld flt
+			fistp intgr
+			} ;
+
+		return (float) intgr ;
+	}
+
 	__inline long long int 
 	llrint (double flt)
 	{	long long int intgr;

msvc_extra_headers/ardourext/misc.h.input

@@ -247,11 +247,6 @@ inline int64_t abs(int64_t val) throw()
 	extern bool operator == (const Type& lhs, const Type& rhs);
 #endif
 
-// round().... Unlike Linux, Windows doesn't seem to support the
-// concept of a system-wide (or programmable) rounding direction.
-// Fortunately, 'round to nearest' seems to be the default action
-// under Linux, so let's copy that until we find out otherwise.
-#define rint(value)  round(value)
 #if !defined(LIBPBD_API) || defined(PBD_IS_IN_WIN_STATIC_LIB)
 extern  double round(double x);