Challenge 2: Classify Shifts as Expected or Unexpected — Possible Solution
====================================================================

(a) A cookie-consent banner appearing at the top of the page 3 seconds
    after load
    -> COUNTS TOWARD CLS (unexpected shift)

    Nothing the user did caused this banner to appear — it shows up on
    its own timer, 3 seconds after load, with no click, tap, or
    keypress triggering it. Because it's injected ABOVE the existing
    page content (at the top), it pushes everything below it down,
    exactly matching the chapter's fourth common cause: "dynamically
    injected content above existing content." The user was likely
    already reading or about to interact with the content that just got
    shoved downward, making this a genuinely disruptive, unexpected
    shift by every definition the chapter gives.

(b) An FAQ answer expanding below a question the user just clicked
    -> DOES NOT COUNT TOWARD CLS (user-initiated shift)

    This shift is a DIRECT, immediate consequence of the user's own
    action — they clicked specifically to expand this content, so its
    appearance and the resulting layout change are expected and
    intentional from the user's own perspective. This matches the
    chapter's explicit carve-out: "a shift caused directly by user
    interaction, like an accordion expanding on click, isn't
    'unexpected' and isn't counted against CLS." The user asked for
    this to happen, so its visual consequences aren't treated as a
    layout-stability defect.

WHY THIS WORKS AS AN ANALYSIS
--------------------------------
The deciding question in both cases is the same: "did the user directly
cause this specific change, close enough in time and clearly enough
tied to their action that they'd expect it?" Case (a) fails this test
entirely — the banner's timing has nothing to do with anything the user
did. Case (b) passes cleanly — the user's click is the direct,
immediate cause of the exact content that then appears. This is exactly
the distinction that separates a genuine CLS problem (content moving
around the user did not ask for) from ordinary, expected interactive
UI behavior (content moving because the user asked it to).