Backfill dependency proxy size in namespace stats

What does this MR do and why?

• Adds a post deploy migration to backfill NamespaceStatistics#dependency_proxy_size which will update the related Namespace::RootStorageStatistics when applicable (if NamespaceStatistics#storage_size changes).
• Moves some EE logic to CE class. NamespaceStatistics now has dependency_proxy_size which is available in CE, so we need to allow for backfilling stats for non-EE now too

Refs https://gitlab.com/gitlab-org/gitlab/-/issues/352853

Query

The query being run in the migration is:

  SELECT dependency_proxy_manifests.group_id FROM dependency_proxy_manifests
  UNION
  SELECT dependency_proxy_blobs.group_id from dependency_proxy_blobs

The cold timings in database-lab are:

Time: 1.238 s
  - planning: 1.229 ms
  - execution: 1.237 s
    - I/O read: 1.039 s
    - I/O write: 0.000 ms

Shared buffers:
  - hits: 3127 (~24.40 MiB) from the buffer pool
  - reads: 1645 (~12.90 MiB) from the OS file cache, including disk I/O
  - dirtied: 137 (~1.10 MiB)
  - writes: 0

https://postgres.ai/console/gitlab/gitlab-production-tunnel-pg12/sessions/9179/commands/32576

The warm timings in database-lab are:

Time: 154.688 ms
  - planning: 0.149 ms
  - execution: 154.539 ms
    - I/O read: 0.000 ms
    - I/O write: 0.000 ms

Shared buffers:
  - hits: 4732 (~37.00 MiB) from the buffer pool
  - reads: 0 from the OS file cache, including disk I/O
  - dirtied: 0
  - writes: 0

https://postgres.ai/console/gitlab/gitlab-production-tunnel-pg12/sessions/9179/commands/32577

The queries being run by the underlying migration / service are unchanged from the original implementation (!55487 (merged)), except they will now also lookup/refresh dependency proxy artifacts as well (added in !79657 (merged)):

SELECT SUM("dependency_proxy_manifests"."size") FROM "dependency_proxy_manifests" WHERE "dependency_proxy_manifests"."group_id" = 184
SELECT SUM("dependency_proxy_blobs"."size") FROM "dependency_proxy_blobs" WHERE "dependency_proxy_blobs"."group_id" = 184 

Processing the backfill for 100 groups locally took 0.8s, so with a batch size of 500, each job will take ~4s.

At the moment, the initial query returns approximately 1388 rows/groups in db-lab (if I'm reading the results correctly), so we can expect something like 3 jobs, with a total time of approx 4.2 minutes

MR acceptance checklist

This checklist encourages us to confirm any changes have been analyzed to reduce risks in quality, performance, reliability, security, and maintainability.

• I have evaluated the MR acceptance checklist for this MR.