Exercise 3: An Untouched Tier's Scenario — Possible Solution
====================================================================

THE SCENARIO
------------------------------
    def scenario_basic_member_gets_no_discount(apply_discount_fn):
        # GIVEN a cart total of $100
        cart_total = 100
        # AND a member with no recognized tier
        membership_tier = "BASIC"
        # WHEN the discount is applied
        result = apply_discount_fn(cart_total, membership_tier)
        # THEN the total is unchanged
        assert result == 100

RESULTS
------------------------------
    Against original apply_discount: PASS
    Against changed apply_discount_v2 (GOLD moved to 15%): PASS

Both versions pass this scenario unchanged.

WHY THIS ONE IS UNAFFECTED
------------------------------
apply_discount's own fallback branch (the implicit "no matching tier"
case, which just returns cart_total unchanged) was never touched by
the GOLD-tier policy change - only the GOLD branch's own multiplier
moved from 0.9 to 0.85. A BDD scenario only fails when the SPECIFIC
behavior it describes actually changes; it says nothing about parts of
the system it doesn't exercise. This scenario, by construction, never
exercises the GOLD branch at all, so it has no way to notice - or care
- that GOLD's own discount changed.

WHY THIS WORKS AS AN ANSWER
------------------------------
This is an important, honest boundary on the chapter's own claim: BDD
scenarios don't provide blanket protection against every possible
change to a system - each scenario only ever verifies the specific
slice of behavior it was written to describe. A full, trustworthy
living specification needs one scenario per meaningful behavior (as
this chapter's own three tiers each got their own), not one scenario
standing in for the whole system. This mirrors Chapter 2's own
fault-localization theme: a scenario that only touches BASIC-tier
behavior can only ever tell you about BASIC-tier behavior, precisely
localized, and nothing about GOLD or PLATINUM.