A good target number for antenna linearity would be one that does not
limit system dynamic range. Our best receivers have a dynamic range of
around 120 dB as measured by the minimum discernable signal on the low
end, and the point where two-tone third order distortion products are
detectable on the high end.

140 dB seems reasonable for an antenna and would theoretically be
measured the same way as receiver dynamic range, though setting up a
noise-free environment, and coupling large distortion-free signals to a
test antenna is a challenge, and is probably one reason we don't see
these measurements. The other reason is that there is good evidence
that a properly built antenna does not limit system dynamic range.
That is, it is very linear in the superposition sense.

By the way, generating new frequencies is not necessarily a violation
of superposition (though it usually is). Consider a system undergoing
a constant Doppler shift.

73,
Glenn AC7ZN