I would think the expectation for accuracy would be 0.1% of the full range since it is a custody transfer meter. A magmeter may pose a problem if no moisture is present and may not be good for this application. If the measurement encounters turbulence and temperature variations, it adds to accuracy problems. Even though installed cost of a true mass flow meter for custody transfer is higher than an annubar/pitot tube, slip stream measurements with scaling up for large pipe lines with Coriolis meter with compensation for temperature and density would be ideal. Simple target meter, vortex shedding meter, turbine meter [Engineered products- insertion type] or time of flight [not Doppler] would be acceptable to keep the accuracy which pays off long term. Budget is always the deciding factor when "pay me now or pay me later" decisions are made.

 Editor's Note: Ram.G.Ramachandran, Systems Research Int'l Inc., contributed to this answer.