Templatize SkPMFloat to support both 1 and 255 biases.

src/opts/SkColorCubeFilter_opts.h

Index: src/opts/SkColorCubeFilter_opts.h
diff --git a/src/opts/SkColorCubeFilter_opts.h b/src/opts/SkColorCubeFilter_opts.h
index 41d2525b5a387a8df89bf017f0e22a59eb3bbae6..fb403a72bdd22fa2f8cd8cc794ecc5ccde636c8f 100644
--- a/src/opts/SkColorCubeFilter_opts.h
+++ b/src/opts/SkColorCubeFilter_opts.h
@@ -51,7 +51,10 @@ void color_cube_filter_span(const SkPMColor src[],
         const int i10 = (colorToIndex[1][g] + colorToIndex[0][b] * dim) * dim;
         const int i11 = (colorToIndex[1][g] + colorToIndex[1][b] * dim) * dim;
 
-        SkPMFloat color(0,0,0,0);
+        // We don't care which SkPMFloat bias we use here, so we'll pick 255,
+        // which makes FromOpaqueColor() and round() a little faster.
+        typedef SkPMFloat<255> PMFloat;
+        PMFloat color(0,0,0,0);
 
         for (int x = 0; x < 2; ++x) {
             const int ix = colorToIndex[x][r];
@@ -61,16 +64,16 @@ void color_cube_filter_span(const SkPMColor src[],
             const SkColor lutColor10 = colorCube[ix + i10];
             const SkColor lutColor11 = colorCube[ix + i11];
 
-            Sk4f  sum = SkPMFloat::FromOpaqueColor(lutColor00) * g0b0;
-            sum = sum + SkPMFloat::FromOpaqueColor(lutColor01) * g0b1;
-            sum = sum + SkPMFloat::FromOpaqueColor(lutColor10) * g1b0;
-            sum = sum + SkPMFloat::FromOpaqueColor(lutColor11) * g1b1;
+            Sk4f  sum = PMFloat::FromOpaqueColor(lutColor00) * g0b0;
+            sum = sum + PMFloat::FromOpaqueColor(lutColor01) * g0b1;
+            sum = sum + PMFloat::FromOpaqueColor(lutColor10) * g1b0;
+            sum = sum + PMFloat::FromOpaqueColor(lutColor11) * g1b1;
 
             color = color + sum * Sk4f((float)colorToFactors[x][r]);
         }
 
         if (a != 255) {
-            color = color * Sk4f(a * 1.0f/255);
+            color = color * Sk4f(a * (1.0f/255));
         }
 
         dst[i] = color.round();