Get the signed mignitude of numbers in interval X, that is the unique number closest to zero for each element.
If X is empty, smig (X) is NaN.
Accuracy: The result is exact.
smig (infsup (-2, -1)) ⇒ ans = -1
See also: @infsup/mig, @infsup/mag, @infsup/inf, @infsup/sup.
Package: interval