VSD - Create partitioned table for counts

For the Value Stream Dashboard overview widget, we want to keep track of various counts in the database.

Proposed schema: value_stream_dashboard_counts

column	type	description
id	bigint	primary key
namespace_id	bigint, foreign key	reference to the namespace record
count	bigint	aggregated count
metric	smallint	which metric is it
recorded_at	timestamp	measurement time

The table should be partitioned by recorded_at (time range) where the partitioning strategy uses the monthly strategy (see WebHookLog as an example). This gives us the opportunity to drop old data if needed. For now, the retain_for option shouldn't be specified (keep all partitions).

Model setup: Analytics::ValueStreamDashboard::Count

Validations:

• namespace_id: required.
• count: required, cannot be less than 0.
• metric: required, enum.
• recorded_at: required.

metric mapping:

{
  projects: 1,
  issues: 2,
  groups: 3,
  merge requests: 4,
  pipelines: 5,
  members: 6
}

Database indexes: only add FK index on namespace_id, the rest will be added when we implement the DB queries.