Fix backup restore and log admin regressions

[cursor]

Bug and impact

• Backup restore UI could not complete because the template submitted fields the active view did not accept and rendered missing backup details.
• Backup delete and log view/delete admin paths could 500 due to missing templates, and backup delete was wired from the list as a GET navigation.
• Log download/delete failed because admin LogEntry audit rows used invalid content_type_id=0.
• Restoring media could wipe/partially replace MEDIA_ROOT after the DB restore had already committed if media copy failed, leaving DB and files inconsistent.

Root cause

The refactored active inventory/views/system/* views drifted from their templates and audit logging used an invalid FK sentinel. Media restore performed destructive filesystem changes after committing the database restore.

Fix

• Aligned restore template/view confirmation fields and backup context usage.
• Added missing backup/log confirmation/view templates and wired list actions to real POST/CSRF or confirmation flows.
• Changed custom admin log entries to nullable content type IDs.
• Staged backup media before DB changes, backed up current media, replaced media inside the DB transaction, and restored old media/re-raised on copy failure so the DB rolls back.

Validation

• python manage.py test inventory.tests.test_views.BackupViewSecurityTest inventory.tests.test_views.LogFileViewTest (9 tests OK; only existing staticfiles warning and expected simulated media failure log)
• python manage.py test inventory.tests.test_sale_status inventory.tests.test_sales_balance_payment inventory.tests.test_services.MemberServiceTest && python manage.py check (11 tests OK; only existing staticfiles warning)