Is it possible to establish a Calibration Base Line (CBL) using only GNSS?
I don't think you can in practice get to the accuracy of an NGS CBL over those distances (typically around a mile and shorter portions). If long-term static sessions get to a few mm accuracy, you would need to do a much longer line and somehow accurately subdivide it.
You would hit diminishing returns on error reduction through redundancy before you attained sufficient accuracy.
The absolute error for GNSS observations is higher than that of total station measurements, while the scale error is smaller. Therefore, for long distances GNSS is more precise than the TS, but for shorter distances - such as those of a calibration baseline, the total stations EDM is more precise than the GNSS.
I think that you could set up something that would be useful with modern GNSS in optimum conditions, but the NGS probably isn't going to let go of using EDMs for this purpose for awhile, yet. For the record, none of the three CBLs in my area would qualify as optimum GNSS conditions. As things progress the difference is becoming smaller. But it is still there.
Leica GS18 (GNSS) specifications:
Leica TS16 (Total Station) EDM specifications: