On some problems concerning best constants for the midpoint and trapezoidal rule