Addendum:

Although it was my intention that the the real and imaginary parts of
the complex numbers are treated separately in the limit-taking, one
could argue that `+inf.0+0.0i` is “reasonably close” in magnitude to
`+inf.0+π/4i`, since both have infinite magnitude. I might change that
to be explicit that it is the real and imaginary parts that must be close.

-- Peter McGoron