Check the application of the frequency mask

@vivien noted that the generated PSD+data plots shows the turn over at half the sampling frequency due to the windowing. This is a known artifact which we handle by applying a frequency mask to the data. We need to

• Check that the frequency mask is applied correctly everywhere
• Not sure the data outside the frequency mask in figures

A neat solution to bopth problems would be to not let the user ever access the unmased data. I.e., replace the ifo.strain_data.frequency_domain_data with ifo.strain_data.frequency_domain_data[ifo.strain_data.frequency_mask] by default. @moritz.huebner @colm.talbot is there any reason not to do this?