Spike: Prepare architecture blueprint for database read model for scan result policies

Time-box: 5 days

Why are we doing this work

In the scope of this Spike, we would like to collaborate as a team to prepare an architecture blueprint for the database read model for Scan Result Policies that we plan to release with Use database read model for merge request appr... (&9971).

The main goals of this redesign are:

• we want to change the current process in https://gitlab.com/gitlab-org/gitlab/-/blob/26952cd8c93a1378873aeed35ca85b6e32becb26/ee/app/workers/concerns/update_orchestration_policy_configuration.rb#L4 not to delete and recreate all related records, but rather update/recreated only affected records,
• reduce the number of queries to the database performed by workers from Security Policies,
• reduce the average duration time of Security::SyncScanPoliciesWorker,

Initial thoughts:

• groupcode review team is working on MVC: Allow group-level MR approval rules for 'A... (&11451) (look at Group approval rules migration (!132651 - merged) to see the context, perhaps we could reuse that),
• we've been discussing with @sashi_kumar that it would be great to have only one record in the database for every policy, then after updating policy.yml, this record would be updated, and necessary changes propagated across GitLab,
• we could see if we can use Gitlab::EventStore for that,

Expected results:

• MR created with an architecture blueprint (similar to Add security policy custom ci blueprint (!137546 - merged)) and initiated discussion with other team members,