fix: eliminate N+1 queries in ContestList scoreboard visibility check

[whlongg]

Description

Type of change: bug fix

What

Eliminates N+1 queries in ContestList when checking scoreboard visibility for hidden-scoreboard contests (SCOREBOARD_HIDDEN / SCOREBOARD_AFTER_CONTEST).

• author_ids / editor_ids: check _prefetched_objects_cache (via getattr) before falling back to through-table queries, so prefetched authors/curators data is reused.
• can_see_full_scoreboard: use Python-side any() over prefetched data instead of .filter(...).exists() per contest.
• ContestList._get_queryset: add view_contest_scoreboard to prefetch_related.
• ContestList.get_context_data: also prefetch contest__view_contest_scoreboard for active participations.

Why

For each contest with a hidden scoreboard, Django was firing 3 extra queries per contest:

1. Contest.authors.through.objects.filter(contest=self) — bypasses prefetch cache
2. Contest.curators.through.objects.filter(contest=self) — same
3. view_contest_scoreboard.filter(id=user.profile.id).exists() — not prefetched at all

All fallback paths are preserved for contexts without prefetch (admin, APIs, single-object views) using getattr(self, '_prefetched_objects_cache', {}).

Fixes #549

How Has This Been Tested?

Automated test — judge/tests/test_contest_list_perf.py:

• Creates 10 contests with SCOREBOARD_HIDDEN, authenticated user.
• Measures SQL query count via connection.queries (DEBUG=True).
• Asserts count ≤ 35. Before fix: 46 queries (FAIL). After fix: 27 queries (PASS).

Direct method benchmark — can_see_own_scoreboard() across multiple scenarios (5 reps each):

+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+
| Scenario                               | Before(time) | Before(q) | After(time) | After(q) | Delta       | Speedup |
+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+
| Visible scoreboard            (N=500)  |      0.9ms   |      0    |      0.9ms  |      0   |     -0.0ms  | 1.02x   |
| Visible scoreboard            (N=1000) |      1.8ms   |      0    |      1.9ms  |      0   | +     0.1ms | 0.95x   |
| Visible scoreboard            (N=2000) |      3.8ms   |      0    |     14.1ms  |      0   | +    10.3ms | 0.27x   |
+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+
| Hidden, no authors/scoreboard (N=500)  |   1123.4ms   |   2001    |    627.8ms  |    500   |   -495.7ms  | 1.79x   |
| Hidden, no authors/scoreboard (N=1000) |   2359.3ms   |   4000    |    859.7ms  |   1000   |  -1499.5ms  | 2.74x   |
+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+
| Hidden, with authors          (N=500)  |    995.1ms   |   2000    |    420.3ms  |    500   |   -574.8ms  | 2.37x   |
| Hidden, with authors          (N=1000) |   1976.6ms   |   4000    |    847.2ms  |   1000   |  -1129.4ms  | 2.33x   |
+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+
| Hidden, with view_scoreboard  (N=500)  |    984.4ms   |   2000    |    417.4ms  |    500   |   -567.0ms  | 2.36x   |
| Hidden, with view_scoreboard  (N=1000) |   1978.3ms   |   4000    |    839.2ms  |   1000   |  -1139.1ms  | 2.36x   |
+----------------------------------------+--------------+-----------+-------------+----------+-------------+---------+

Before fix: 4 queries/contest (2× through-table + exists + has_completed_contest).
After fix: 1 query/contest (only has_completed_contest remains, out of scope).

# Automated regression test
python manage.py test judge.tests.test_contest_list_perf
# [10 hidden-scoreboard contests] query count: 27 (budget: 35)
# Ran 1 test in 0.475s — OK

• judge.tests.test_contest_list_perf — query count bounded (27 ≤ 35, before fix was 46)

Checklist

• I have explained the purpose of this PR.
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the README/documentation
• Any dependent changes have been merged and published in downstream modules
• Informed of breaking changes, testing and migrations (if applicable).

By submitting this pull request, I confirm that my contribution is made under the terms of the AGPL-3.0 License.