To clarify things:
Are you are asking why the authors of the paper : "Halloin et al, ‘Long term characterization of voltage references’,submitted to IEEE trans. on Instr. & Meas." chose to test only those references in paper and not LZT1000, LM399 or LTFLU?
Answer is : how would we know, if it doesn't say anywhere? Send them an E-mail, maybe they will answer.
If I had to speculate, it is either that there is low power requirement (heated references need lot of power), or if you look at referenced project, it seems this thing should fly to space. Maybe these references mentioned are available in space certified version?
LTZ1000 is actually very rad-hard, problem is that surrounding circuitry has to be made so too. So flying LTZ1000 to space is probably about certifying whole LTZ1000 reference module as a unit. SAme with LM399 and LTFLU..
But as I said only authors know what was selection criteria in the first place.