Move after_commit hook from AntiAbuse::TrustScore to a worker

What does this MR do and why?

Related to https://gitlab.com/gitlab-org/modelops/anti-abuse/team-tasks/-/issues/849

Move after_commit hook from AntiAbuse::TrustScore to a worker

To avoid deadlock conditions when many trust scores are created for a single user, the after_commit hook in AntiAbuse::TrustScore is moved to a sidekiq job that will only execute the cleanup once every 5 minutes for a given user.

MR acceptance checklist

Please evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

Screenshots or screen recordings

Screenshots are required for UI changes, and strongly recommended for all other merge requests.

Before	After

How to set up and validate locally

Numbered steps to set up and validate the change are strongly suggested.

1. Apply a patch to lower the trust score limit and cache timeout for easier testing.

   diff --git a/app/models/anti_abuse/user_trust_score.rb b/app/models/anti_abuse/user_trust_score.rb
   index a2add9d4c9e3..9bd4973f06bc 100644
   --- a/app/models/anti_abuse/user_trust_score.rb
   +++ b/app/models/anti_abuse/user_trust_score.rb
   @@ -2,7 +2,7 @@
   
    module AntiAbuse
      class UserTrustScore
   -    MAX_EVENTS = 100
   +    MAX_EVENTS = 3
        SPAMCHECK_HAM_THRESHOLD = 0.5
   
        def initialize(user)
   diff --git a/app/workers/anti_abuse/trust_score_cleanup_worker.rb b/app/workers/anti_abuse/trust_score_cleanup_worker.rb
   index 77b1289fae63..4a30b5441902 100644
   --- a/app/workers/anti_abuse/trust_score_cleanup_worker.rb
   +++ b/app/workers/anti_abuse/trust_score_cleanup_worker.rb
   @@ -18,7 +18,7 @@ def perform(user_id, source)
          return if Rails.cache.exist?(cache_key)
   
          AntiAbuse::UserTrustScore.new(user).remove_old_scores(source)
   -      Rails.cache.write(cache_key, true, expires_in: 5.minutes)
   +      Rails.cache.write(cache_key, true, expires_in: 1.minutes)
        end
      end
    end

2. Tail sidekiq logs to observe that the workers are executing.

   tail -f log/sidekiq.log | grep -i trustscore

3. Execute the Abuse::TrustScoreWorker 4 times. You should see the Abuse::TrustScoreWorker and the AntiAbuse::TrustScoreCleanupWorker being queued.

   user = User.last
   Abuse::TrustScoreWorker.perform_async(user.id, Enums::Abuse::Source.sources[:spamcheck], rand(0.0...1.0))
   Abuse::TrustScoreWorker.perform_async(user.id, Enums::Abuse::Source.sources[:spamcheck], rand(0.0...1.0))
   Abuse::TrustScoreWorker.perform_async(user.id, Enums::Abuse::Source.sources[:spamcheck], rand(0.0...1.0))
   Abuse::TrustScoreWorker.perform_async(user.id, Enums::Abuse::Source.sources[:spamcheck], rand(0.0...1.0))

4. Check the count of abuse trust score records for the user. There should be 4 records since the cleanup worker was executed before the limit was exceeded.

   user.abuse_trust_scores.where(source: Enums::Abuse::Source.sources[:spamcheck]).count

5. Wait for 1 minute and execute the Abuse::TrustScoreWorker once more. At this point, the cache should have expired and the count of trust score records should be reduced to 3.