So use a really bad DVM
The pairs for base metal thermocouples are chosen to give a relatively
high voltage for a given temperature difference (among other things).
Typical connection material pairs are something like 5:1~10:1 better,
and you'll very seldom see anything other than a "0.0" mV if you go
around probing random bits of metal that have just been handled by
30°C fingers. A DC shunt will be made symmetrical in part so the
substantial self-heating at rated current won't cause thermocouple
voltages to affect the reading. Immediately when the current is shut
off, the meter should go to 0. If it doesn't, then the reading can be
corrected by that factor, but it will not be a problem with such a
setup and a 100uV resolution meter.
If you want to calibrate a really high current shunt (not just check
it) at a current orders of magnitude less than the rated current, then
such things would come into play (and perhaps you'd be using a meter
with 100nV resolution rather than 100uV), but the OP just wants to
assure himself that it's actually 50mV at rated current as marked.
We've supplied many, many, high current (up to 15,000A) DC measurement
systems using such shunts, BTW. They're usually between 50mV and 150mV
output at rated current.
Best regards,
Spehro Pefhany