The TF looks good. But the noise measurement is obviously limited by the SR785 noise. We need a preamp, which is only for the purpose of the measurement. It has to have the input reffered noise about a factor of a few better than the noise predicted by Zero. At high frequency, probably we will be able to use SR560. With this low noise level, probably we can just use the flat gain of 100 for the SR560 setting. This will give you the input referred noise (of the preamp) of ~4nV/rtHz at kHz band. Note that the gain needs to be larger than 100 to have low noiseness of SR560.
I think this is a solid measurement.