libcxx properly defines overriden fmin() in math.h.

Unfortunately, that one has exception specifications, so redefining it
does not go so well.
Simply rename it so that the conflict never happens.

graphics/djvulibre/patches/patch-libdjvu_DjVuPalette_cpp

@@ -0,0 +1,11 @@
+$OpenBSD: patch-libdjvu_DjVuPalette_cpp,v 1.1 2017/04/20 01:40:33 espie Exp $
+--- libdjvu/DjVuPalette.cpp.orig	Thu Apr 20 03:30:12 2017
++++ libdjvu/DjVuPalette.cpp	Thu Apr 20 03:38:11 2017
+@@ -98,6 +98,7 @@ inline unsigned char 
+ umin(unsigned char a, unsigned char b) 
+ { return (a>b) ? b : a; }
+ 
++#define fmin myfmin
+ inline float 
+ fmin(float a, float b) 
+ { return (a>b) ? b : a; }