Use filesystem usage for incomplete snapshots instead of trusting potentially
stale metadata, and expose unique non-hardlinked data totals for completed
snapshots.
Update dashboard and host storage summaries so incomplete data is visible and
complete snapshot totals distinguish allocated and unique data.
- Fall back to filesystem measurement when snapshot storage metadata is missing.
- Prefer `data/` inside snapshot directories so incomplete snapshot metadata/log files are not counted as backup data.
- Add stats and dashboard rendering coverage for incomplete snapshots without recorded storage metadata.
## Tests
- .venv/bin/python manage.py test src.pobsync_backend.tests.test_stats_summary src.pobsync_backend.tests.test_views.ViewTests.test_dashboard_host_cards_measure_incomplete_data_without_snapshot_metadata --verbosity 2
- .venv/bin/python manage.py check
- git diff --check
- .venv/bin/python manage.py test src.pobsync_backend --verbosity 2
Closes#62
Aggregate snapshot storage metadata by snapshot kind so operators can see
scheduled, manual, incomplete, and total backup data.
Surface the totals per host and across all hosts on the dashboard, using
allocated snapshot size from recorded backup metadata without rescanning
backup storage.
