Use frequency offsets as coupling for optical pathlength noises

We currently use an approach similar to LISANode: optical pathlength noises are expressed in fractional frequency deviations, and they are coupled to our simulated laser beams frequencies with the laser central frequency as the coupling coefficient. So we approximate the laser frequency with the laser central frequency, making an error of 10 MHz / 100 THz ~ 1E(7-14) = 1E-7 for the carriers, and 1 GHz / 100 THz ~ 1E(9-14) = 1E-5 for the sidebands.

We suggest to use central_freq + offsets as the coupling coefficient. This shouldn't add much complexity to the simulation, and reduce the errors to around 1 Hz / 100 THz ~ 1E-14 for both carriers and sidebands.