Speed up JudgeHandler's processing of problem list

[hieplpvip]

Description

Type of change: improvement

What

Speed up JudgeHandler's processing of problem list

Why

OJ has more than 40k+ problems now. Judges sometimes stall waiting for bridged to process the problem list.

When test data for any problem changes, judges send the codes of ALL judge-able problems to bridged, which then maps these codes to primary keys:

judge.problems.set(Problem.objects.filter(code__in=list(self.problems.keys())))

This is extremely slow because:

1. This retrieves all fields, not just primary keys.
2. This creates a model instance for every returned row just to be discarded right after set runs.

Mitigated by only querying the primary keys as values list.

A proper fix for this could be caching the mapping, but it's not straightforward as problems can be renamed.

How Has This Been Tested?

On my PC, I simulate processing all problems:

_problems = Problem.objects.values_list('code', flat=True).all()
_problems = list(zip(_problems, [0] * len(_problems)))
self._problems = _problems

Previously, this takes 0.351s. After this patch, it only takes 0.129s :)

Be aware that my PC is faster than our server, and I only run 1 judge instead of 4. The processing time would be further multiplied by the number of judges.

Checklist

• I have explained the purpose of this PR.
• I have performed a self-review of my own code

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