That’s kind of the beauty here, you only get a relative error (0.069%) and it’s mostly insignificant to the point of being smaller than your fabrication tolerances. This should also be the case for fibre optics. The error can’t really add up, you scale everything to a multiple of the wavelength and you calculate the wavelength using the speed of light.