inject_signal_into_gwpy_timeseries calculates SNR with wrong PSD

The interferometers initialized in this line, called from here in bilby_pipe are never updated to include the PSD actually being used for the analysis. This doesn't affect the injection procedure, but it means that the SNRs are calculated using the default PSDs for each interferometer, so the printed SNRs in the log files are wrong. One solution would be for the function to be modified to take in the PSDs as arguments in order to calculate the SNRs correctly, or otherwise it should just not return the calculated SNRs at that stage, and they can be calculated later in the bilby_pipe_generation.