Fix mixed-type usage for modf

When building Qt with a non-double qreal type, i.e.
QT_COORD_TYPE=float, the C function modf rejects the float type since it
is declared as:

        float modff(float, float*)
        double modf(double, double*)

Fix this by porting to std::modf, which provides overloads for both
FP types.

Amends 91f502a7d65d76f4a376903c4058385a62dad277.

Change-Id: Ic1de2f3d8e75a9594dbbd0fa92e63505dae6f1b4
Reviewed-by: Marc Mutz <marc.mutz@qt.io>

src/gui/painting/qpainterpath.cpp

@@ -22,6 +22,8 @@
 #include <private/qstroker_p.h>
 #include <private/qtextengine_p.h>
 
+#include <cmath>
+
 #include <limits.h>
 
 #if 0
@@ -3228,7 +3230,7 @@ QPainterPath QPainterPath::trimmed(qreal f1, qreal f2, qreal offset) const
 
     if (offset) {
         qreal dummy;
-        offset = modf(offset, &dummy); // Use only the fractional part of offset, range <-1, 1>
+        offset = std::modf(offset, &dummy); // Use only the fractional part of offset, range <-1, 1>
 
         qreal of1 = f1 + offset;
         qreal of2 = f2 + offset;