Lube, or no lube?
I imagine that this reliability criterion is more like "fails the
resistance spec" because drift would require them to fine
measure, and track data, on a whole lot of pins that have
minuscule initial resistance values.
Or if you were a smart guy you might rig up a daisy-chain
along the mating pair and measure -that-, maybe look for
a delta failure, but now you have no idea whether it was
one pin went grossly bad, or all of 'em a little bit, without
some failure analysis.
I'd call this a good question for the mfr's applications
engineering group. They might even put their hands on
reliability test data that shows R vs matings and goes
further out, maybe shows individual samples so you
can get a feel for whether this is defect or whole
population movement.