Improve admin backend list performance

[HCDD1]

What this PR does

Improves the admin backend user list performance by moving filtering and pagination into the database and batching storage usage aggregation for the users on the current page.

Fixes #2

/claim #2

How to use

No UI or API contract changes are required. The existing admin users endpoint should behave the same, but avoid loading every user and every user's documents for each list request.

How it works

• Adds a paged admin user query with optional search and admin/regular filtering.
• Adds a batched document storage stats query grouped by owner ID.
• Replaces per-user document loading with one aggregate query for the current page.
• Uses countByAdminTrue() for system info instead of loading all users into memory.

How I tested it

• Added unit coverage for the paged admin user list and batched storage stats behavior.
• Added unit coverage to ensure system info counts admins without calling findAll().

Local test execution is currently blocked on this machine because JAVA_HOME is not configured and no java executable is available in PATH.