Indeed, tail-call optimization is not part of the standard, unlike Scheme.

The point of the article is that while the Common Lisp standard is not really focused on functional programming, nothing prevents the implementions (and libraries) of today to be so.