Exercise 2: Two More Tried Values — Possible Solution
====================================================================

EXTENDED DATA
------------------------------
    threshold_results = {
        0.3:  {'recall': 0.95, 'false_positive_rate': 0.40},
        0.5:  {'recall': 0.80, 'false_positive_rate': 0.15},
        0.55: {'recall': 0.75, 'false_positive_rate': 0.10},
        0.65: {'recall': 0.60, 'false_positive_rate': 0.05},
        0.7:  {'recall': 0.55, 'false_positive_rate': 0.04},
    }
    exploration_order = [0.3, 0.5, 0.55, 0.7, 0.65]

RESULTS
------------------------------
    tried 0.3:  fpr=0.40 -> current best: 0.3
    tried 0.5:  fpr=0.15 -> current best: 0.5   (rewrite 1)
    tried 0.55: fpr=0.10 -> current best: 0.55  (rewrite 2)
    tried 0.7:  fpr=0.04 -> current best: 0.7   (rewrite 3)
    tried 0.65: fpr=0.05 -> current best: 0.7   (no rewrite - 0.65 is worse than 0.7)

    Total test rewrites with 5 tried values: 3
    (the chapter's own 3-value exploration needed 2 rewrites)

WHY THIS WORKS AS AN ANSWER
------------------------------
Adding more explored values increased the rewrite count from 2 to 3 -
not because the underlying problem changed, but because trying more
candidate thresholds naturally means more chances to find a better one
partway through. The one value that DIDN'T cause a rewrite (0.65,
tried last) illustrates the flip side: exploration doesn't always
improve on the current best, and a TDD-first test would have been
correct not to change on that particular attempt. This confirms the
chapter's own point scales with how much genuine exploration a task
actually requires - the more open-ended the search, the more a
test-first assertion has to be rewritten along the way.