A zero-ohm jumper constructed as a SMD resistor,
will have the same reliability attributes / issues as
any other so-constructed component. It is no worse
risk than the rest.
Smaller components are better since the thickness
is likely similar and "moment arm" is smaller. Since
there is zero resistance there is zero power dissipation
and no reason for a large body other than rated
terminal current (internal or external, land/trace size?).
If you want an early answer then you want accelerated
testing (you do not have to wait months for, say, 1K
temp cycles). But I'm guessing that along with "late
before you start" product development paradigm,
comes "spend no money" (because time is money and
having neither, seems to be a theme here).
Really, dude, you need a better class of customer
/ employer. It's just one thing after another and
none of it good.