Use std:: variants for sin(), cos() and sqrt().

They should have overloads for given type, so we're not using doubles
when only float is needed.

src/Math/Matrix4.h

@@ -75,8 +75,8 @@ template<class T> class Matrix4: public Matrix<4, T> {
         static Matrix4<T> rotation(T angle, const Vector3<T>& vec) {
             Vector3<T> vn = vec.normalized();
 
-            T sine = sin(angle);
+            T sine = std::sin(angle);
-            T cosine = cos(angle);
+            T cosine = std::cos(angle);
             T oneMinusCosine = T(1) - cosine;
 
             T xx = vn.x()*vn.x();

src/Math/Vector.h

@@ -90,7 +90,7 @@ template<size_t size, class T> class Vector {
          * @todo optimize - Assume the vectors are normalized?
          */
         inline static T angle(const Vector<size, T>& a, const Vector<size, T>& b) {
-            return acos(dot(a, b)/(a.length()*b.length()));
+            return std::acos(dot(a, b)/(a.length()*b.length()));
         }
 
         /** @brief Default constructor */
@@ -248,7 +248,7 @@ template<size_t size, class T> class Vector {
          * @see lengthSquared()
          */
         inline T length() const {
-            return sqrt(dot(*this, *this));
+            return std::sqrt(dot(*this, *this));
         }
 
         /**