This value is chosen to be a few powers of 2 less than the
number of bits available in the double representation (i.e. 53).
This should allow enough extra precision for simple computations to be correct,
at least for comparison purposes.
Computes whether the interval min, max is effectively zero width.
I.e. the width of the interval is so much less than the
location of the interval that the midpoint of the interval cannot be
represented precisely.
/