Rethinking resolution estimation in fluorescence microscopy: from theoretical resolution criteria to super-resolution microscopy