[Tracking] Admin Computing Units Dashboard

### Feature Summary

Add an Admin -> Computing Units view to the admin dashboard so admins can monitor and reclaim computing units cross all users. Follow-up to discussion #6322

### Motivation
On resource-limited Kubernetes deployments, CU creation frequently fails because the cluster is full. Today an admin has no in-app tooling. They must SSH in, run `kubectl` to list CU pods, and cross-reference the `workflow_computing_unit` table to find owners before manually deleting pods.

### Related
Discussion #6322
Discussion #6264 / PR #6046 (auto-terminating idle CUs) - coordinate to avoid duplicate termination logic and double notifications.

### Proposed Solution or Design

1. Monitoring - table of all CUs: name, owner, type, status, creation time, resource specs
2. Management - admin terminate action reusing the existing termination path.
3. Cluster overview - total vs. allocated vs. available CPU / memory / GPU.

### Affected Area

Deployment / Infrastructure