Use decimal instead of float in WVec.LerpQuadratic

To reduce desync risk (without introducing overflow risk).

OpenRA.Game/WVec.cs

@@ -82,8 +82,8 @@ namespace OpenRA
 				return ret;
 
 			// Add an additional quadratic variation to height
-			// Uses fp to avoid integer overflow
+			// Uses decimal to avoid integer overflow
-			var offset = (int)((float)(b - a).Length * pitch.Tan() * mul * (div - mul) / (1024 * div * div));
+			var offset = (int)((decimal)(b - a).Length * pitch.Tan() * mul * (div - mul) / (1024 * div * div));
 			return new WVec(ret.X, ret.Y, ret.Z + offset);
 		}