Draw from flat distribution of the SPE charge.

[fschlueter]

As far as I understand this is an approximation (also indicated by the comment)

                    # ToDo: figure out actual chage distribution of pulses!
                    pulse_charges = self.random_service.uniform(
                        0.25, 1.75, size=num_pe)

[mhuen]

Yes. If you really want more accurate simulations with egenerator, then one might consider to train a model to simulate MCPE instead of reco pulses. The above would then simply be 1 photon and detector simulation would handle things afterwards.