Premier League Predictor: Django & MySQL — Chapter 6, Exercise 3
====================================================

TASK
Enter an incorrect result for a real fixture with at least two
predictions already recorded via enter_result, confirm their
points_awarded values, then call enter_result again with the corrected
result and confirm every prediction's points_awarded genuinely changed
via bulk_update() rather than only being scored the first time.

SOLUTION

Using fixture_id 42 with two predictions already recorded (a user
prediction of 2-1, an expert prediction of 1-1):

1. Enter a deliberately wrong first result:

   POST /fixtures/42/result/
   {"home_score": 1, "away_score": 1}

   Inside enter_result, this sets fixture.home_score = 1,
   fixture.away_score = 1, status = "played", saves the fixture, then
   builds a list of both Prediction rows and computes points_awarded
   for each before calling bulk_update().

2. Call GET /fixtures/42/predictions/ and check points_awarded:

   - user (2-1): _outcome(2, 1) is "home_win"; the entered result (1,
     1) is a draw. No exact match, no outcome match -> points_awarded
     = 0.
   - expert (1-1): exact match against the entered (1, 1) result ->
     points_awarded = 40.

3. Now correct the result to the real one:

   POST /fixtures/42/result/
   {"home_score": 2, "away_score": 1}

   enter_result runs exactly the same way a second time: it re-fetches
   both Prediction rows fresh from the database, recomputes
   points_awarded for both against the new (2, 1) result, and calls
   bulk_update(predictions, ['points_awarded']) again.

4. Call GET /fixtures/42/predictions/ again and check points_awarded:

   - user (2-1): now an exact match against the corrected (2, 1)
     result -> points_awarded = 40.
   - expert (1-1): a draw prediction against a corrected home-win
     result -> different outcomes -> points_awarded = 0.

Both predictions' points_awarded values genuinely flipped between the
two calls — the user went from 0 to 40, the expert went from 40 to
0 — confirming that enter_result really does rebuild the full list of
predictions and rescore every one of them from scratch on every call,
and that bulk_update() writes the corrected values back to the
database rather than the first call's values silently persisting.

WHY THIS WORKS AS AN ANSWER
----------------------------
It walks through a real before-and-after sequence using two
predictions deliberately chosen so the correction flips which one
scores well (rather than one where both happen to end up scored the
same either way), making the change genuinely visible and verifiable
in both directions, and explicitly ties the outcome back to
enter_result re-running its full loop and bulk_update() call on every
invocation, not just the first.