This issues are known for a long time if you read the posts .
Autorange is slow .
50ohm range may be redundant because they did it wrong . The protection thermistor PTC3 little drift with temperature is responsible for getting different results all the time . Maybe there is some self heating from the measurement current itself , not just the internal temperature variation . 2 wire low ohm measurement is not that simple to do it right .
Later edit
I played a little with the board out and ... it's not the thermistor heating up ... The problem is that the measurement current is low ( maybe too low ) and it's using the lowest range possible with 10 times internal amplification ( in HY3131 configuration is like that ) . Every solder joint is a thermocouple at this low level . That's why it drifts all the time and can't be stable .
With the thermistors shunted , any solder joint I heat up ( in input zone ) results in massive 1000 counts drift . Thats 1 ohm . In other ranges the drift is lower , as expected . What's impossible to say at this point is if they use HY3131 correctly in 50ohm range as in configuration pdf document .