I don't know. Maybe just use what R&S uses for the other standards. I don't think this is going to affect your measurement by 0.1 dB since the difference is 0.0033 dB between you two values. The larger contributions will be isolation for an S21 measurment, source match for the S11 measurement, and repeatability of connectors. Agilent has a measurement uncertainty spreadsheet for VNAs that you may be able to use. Maybe R&S has similar. Then again, are you looking for a 0.1 dB repeatability or 0.1 dB accuracy to a know loss standard. The repeatability is probably easier to get.