Premier League Predictor: FastAPI & Redis — Chapter 8, Exercise 2
=================================================================

TASK
Use the script to make the user and the expert finish on the same points
but with different correct-score counts, then make three sources tie on
both keys. Report the leaderboard for each case and say what decided it.

SOLUTION

Case 1 - tie on points, different correct-score counts
  Fixture 1: user 40 points, expert 0, AI 0.
  Fixtures 2-5: expert 10 points each.
  totals:  user 40, expert 40
  records: user:scores 1, expert:results 4, expert:scores 0
  Leaderboard: user (-4000001), expert (-4000000), ai (0), guest (0)
  The user is ahead: both have 4000 hundredths x 1000, but the user's one
  correct score adds 1 to the composite (which makes the score more
  negative, so it sorts first).

Case 2 - three sources tied on both keys
  Fixture 1: user 10, expert 10, AI 10 (correct results, no correct scores).
  Leaderboard: ai (-1000000), expert (-1000000), user (-1000000), guest (0)
  Nothing separates them, so Redis orders equal scores by member name,
  ascending: ai, expert, user. This is an arbitrary fallback, not a rule
  of the competition; the guest, with no points, is last.

WHY THIS WORKS AS AN ANSWER
----------------------------
Both orderings come from real runs of the chapter's script, each attributed
to the exact part of the composite that decided it, and the full tie is
identified as an arbitrary fallback rather than passed off as meaningful.