feat: interface PIN, bulk disposition tagging, Google Drive delta scan, OCR memory fixes

- Interface PIN: optional session-level auth gate for the main scanner UI
  (Settings → Security → Interface PIN). Salted SHA-256 in config.json,
  rate-limited (5 attempts/5 min per IP). /view and viewer auth exempt.
  New /login page, before_request hook, GET/POST/DELETE /api/interface/pin,
  POST /api/interface/pin/verify, POST /api/interface/logout.

- Bulk disposition tagging: Select mode (filter bar "Vælg" button) reveals
  per-card checkboxes. Bulk tag bar at bottom of grid; POST /api/db/disposition/bulk.
  Disposition stats bar (total · unreviewed · retain · delete · % reviewed)
  updates after every save.

- Google Drive delta scan: uses Drive Changes API when delta is enabled.
  Per-user token stored as gdrive:{email} in delta.json. Load-then-merge
  save avoids racing with concurrent M365 token writes.

- PDF OCR OOM fix: render one page at a time with convert_from_path
  (first_page=N, last_page=N). Added _ocr_mem_ok() psutil guard (500 MB
  threshold) before each page render across scan_pdf, redact_fitz_pdf,
  redact_pdf.

- Email test message translation fix: routes/email.py returns structured
  {ok, method, recipients} instead of a hardcoded English string;
  scheduler.js builds the translated message client-side.

- Docs: CHANGELOG, README, TODO, MANUAL-EN, MANUAL-DA all updated.
  Lang files (en/da/de) extended with bulk, interface PIN, and SMTP keys.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

routes/CLAUDE.md

@@ -14,6 +14,9 @@ All three scan engines must include `"source": "m365"` / `"google"` / `"file"` i
 ## Circular import prohibition
 `scan_engine.py` and `gdpr_scanner.py` must not import each other. `scan_engine` imports from `sse`, `checkpoint`, `app_config`, `cpr_detector`; `gdpr_scanner` imports scan functions from `scan_engine`.
 
+## `_scan_bytes` injection
+`scan_engine.py` declares stub versions of `_scan_bytes` / `_scan_bytes_timeout` at module level. `gdpr_scanner.py` replaces them with the real `cpr_detector` implementations at startup. `routes/google_scan.py` pulls them from `gdpr_scanner` via `__getattr__`. Never import these directly in blueprint or engine modules — that breaks the circular-import barrier.
+
 ## Gotchas
 
 - **`_load_settings()` return** — does NOT include `file_sources`. Returns only: sources, user_ids, options, retention_years, fiscal_year_end, email_to.