handle negative intermediate value in rgb to hsv conversion

Python's `%` and C's `fmod` use a slightly different strategy. So to match Python, we need to implement it ourselves. It can also be simplified some due to the second parameter being 1.

The starting hue calculation has also been simplified by inlining some values.

src/libImaging/Convert.c

@@ -310,40 +310,45 @@ rgb2bgr24(UINT8 *out, const UINT8 *in, int xsize) {
 }
 
 static void
-rgb2hsv_row(UINT8 *out, const UINT8 *in) {  // following colorsys.py
-    float h, s, rc, gc, bc, cr;
-    UINT8 maxc, minc;
-    UINT8 r, g, b;
-    UINT8 uh, us, uv;
-
-    r = in[0];
-    g = in[1];
-    b = in[2];
-    maxc = MAX(r, MAX(g, b));
-    minc = MIN(r, MIN(g, b));
-    uv = maxc;
+rgb2hsv_row(UINT8 *out, const UINT8 *in) {
+    // based on Python's colorsys module
+
+    const UINT8 r = in[0];
+    const UINT8 g = in[1];
+    const UINT8 b = in[2];
+
+    const UINT8 maxc = MAX(r, MAX(g, b));
+    const UINT8 minc = MIN(r, MIN(g, b));
+
+    UINT8 uh, us;
+    const UINT8 uv = maxc;
+
     if (minc == maxc) {
         uh = 0;
         us = 0;
     } else {
-        cr = (float)(maxc - minc);
-        s = cr / (float)maxc;
-        rc = ((float)(maxc - r)) / cr;
-        gc = ((float)(maxc - g)) / cr;
-        bc = ((float)(maxc - b)) / cr;
+        const UINT8 color_range = maxc - minc;
+        double h;
+
+        const double cr = (double)color_range;
         if (r == maxc) {
-            h = bc - gc;
+            h = (g - b) / cr;
         } else if (g == maxc) {
-            h = 2.0 + rc - bc;
+            h = 2.0 + (b - r) / cr;
         } else {
-            h = 4.0 + gc - rc;
+            h = 4.0 + (r - g) / cr;
         }
-        // incorrect hue happens if h/6 is negative.
-        h = fmod((h / 6.0 + 1.0), 1.0);
 
-        uh = (UINT8)CLIP8((int)(h * 255.0));
-        us = (UINT8)CLIP8((int)(s * 255.0));
+        // the modulus operator in Python does not exactly match fmod in C
+        // https://stackoverflow.com/a/3883019/3878168
+        // "h = (h/6.0) % 1.0" in Python can be computed as:
+        h = h / 6.0;
+        h = h - floor(h);
+
+        uh = (UINT8)(255.0 * h);
+        us = 255 * color_range / maxc;
     }
+
     out[0] = uh;
     out[1] = us;
     out[2] = uv;