Confusing behaviour of injection

I just want to highlight some potentially odd behaviour when injecting code that may be confusing. Here is a script to produce an IFO object with a simulated signal in noise

import peyote
import matplotlib.pyplot as plt
import numpy as np

time_duration = 1.
sampling_frequency = 4096.

simulation_parameters = dict(
    mass_1=36.,
    mass_2=29.,
    spin_1=[0, 0, 0],
    spin_2=[0, 0, 0],
    luminosity_distance=410.,
    inclination_angle=0.,
    waveform_phase=0.,
    waveform_approximant='IMRPhenomPv2',
    reference_frequency=50.,
    ra=0,
    dec=1,
    geocent_time=0,
    psi=1
    )

source = peyote.source.BinaryBlackHole(
    'BBH', sampling_frequency, time_duration)
hf_signal = source.frequency_domain_strain(simulation_parameters)


IFO = peyote.detector.H1
hf_noise, frequencies = IFO.power_spectral_density.get_noise_realisation(
    sampling_frequency, time_duration)
plt.loglog(frequencies, hf_noise)
IFO.set_data(frequency_domain_strain=hf_noise)
IFO.inject_signal(source, simulation_parameters)
IFO.set_spectral_densities(frequencies)
IFO.whiten_data()

plt.loglog(frequencies, hf_noise)
plt.show()

I see three confusing points

1. hf_noise is returned from IFO.power_spectral_density.get_noise_realisation and then immediately set. Can we have a wrapper which does this all in one?

2. It is opaque to a user what set_spectral densitiesandwhiten_dataare going (I'm not sure I understand at least). If these are always required, can they be done in theinit` or some other automatic way?

3. You'll notice I've plotted hf_noise twice - see the two plt.loglog calls. These produce the ba ckground noise by itself, then the background noise + signal. This is quite perplexing from a users point of view. Can it be made more explicit?

Note: for 3, I think (but I'm unsure) that this is one of those cases where you don't have the data, but a pointer.

Tagging @paul-lasky @colm.talbot @nikhil.sarin as people who may know more about the detectors internals.