In the Y6 testing, we discovered (due to a different bug that was fixed in #136) that the weight adjustment we do to bring nominal S/N values down to the requested max_snr value is adding a bit too much noise. If calculateSNR is rerun after this step, the measured SNR is often about 10% lower. Around 90 when the requested max_snr=100. Apparently there is something wrong in the math there, which should be fixed.
In the Y6 testing, we discovered (due to a different bug that was fixed in #136) that the weight adjustment we do to bring nominal S/N values down to the requested max_snr value is adding a bit too much noise. If calculateSNR is rerun after this step, the measured SNR is often about 10% lower. Around 90 when the requested max_snr=100. Apparently there is something wrong in the math there, which should be fixed.