Exercise 1: Four Levels of Nesting — Possible Solution
====================================================================

THE TEST
------------------------------
    naive_extract("<div><div><div><div>Deepest</div></div></div></div>", "div")

RESULT
------------------------------
    ['<div><div><div>Deepest']

THE PATTERN
------------------------------
    2 levels deep: "<div><div>Inner</div></div>"
      -> captures '<div>Inner'                      (1 stray opening tag)

    3 levels deep: "<div><div><div>Deep</div></div></div>"
      -> captures '<div><div>Deep'                  (2 stray opening tags)

    4 levels deep: "<div><div><div><div>Deepest</div></div></div></div>"
      -> captures '<div><div><div>Deepest'          (3 stray opening tags)

The rule: N levels of nesting leaks exactly N-1 stray opening tags into
the "captured" result.

WHY THIS WORKS AS AN ANSWER
------------------------------
The regex always starts matching at the FIRST <div> in the string and
scans forward, capturing every raw character until it hits the first
literal </div> -- with zero awareness that any of the div tags along
the way were themselves opening tags. Every additional level of
nesting adds exactly one more <div> opening tag between the true
outer tag and the first closing tag the regex will find -- and since
the regex has no counter tracking how many of those openings it has
already passed, every single one of them gets swept into the
"content" it reports. The relationship is linear and exact: it isn't
that deeper nesting makes the regex "more confused" in some vague
sense -- it fails by precisely one additional leaked tag per
additional level, which is exactly what you'd expect from a tool that
is counting zero levels of anything at all.