Get the midpoint of interval X.
If X is empty, mid (X) is NaN.
If X is entire, mid (X) is 0.
If X is unbounded in one direction, mid (X) is positive
or negative realmax ().
Accuracy: The result is rounded to the nearest floating point number and
may thus be exact or not. However, it is guaranteed that the interval
X is tightly enclosed by
[mid (X) - rad (X), mid (X) + rad (X)].
mid (infsup (2.5, 3.5)) ⇒ ans = 3
See also: @infsup/inf, @infsup/sup, @infsup/rad.
Package: interval