fix(ocr,gitignore): fix charmap encoding in OCR workers and ignore SQLite WAL files

- Fix UnicodeEncodeError in OCR worker subprocesses on Windows: spawned processes
  with 'spawn' context inherit charmap (CP1252) encoding, causing crashes when
  printing Romanian diacritics (ș, ț, ă) from OCR-extracted text.
  Fix: reconfigure sys.stdout/sys.stderr to UTF-8 in _worker_initializer().

- Add *.db-wal and *.db-shm to .gitignore for SQLite WAL mode auxiliary files.

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

.gitignore

@@ -519,6 +519,8 @@ secrets-backup/**/.env.*
 # ============================================================================
 # Backend unified data directories (cache, receipts, telegram)
 backend/data/cache/*.db
+backend/data/cache/*.db-wal
+backend/data/cache/*.db-shm
 backend/data/receipts/*.db
 backend/data/telegram/*.db
 backend/data/receipts/uploads/*

backend/modules/data_entry/services/ocr/ocr_worker_pool.py

@@ -367,6 +367,17 @@ def _worker_initializer() -> None:
 
     Total warmup time = max(engine_times) instead of sum(engine_times).
     """
+    import io
+
+    # Fix stdout/stderr encoding in spawned worker process.
+    # On Windows, 'spawn' context creates a subprocess with charmap (CP1252) encoding.
+    # Romanian diacritics (ș, ț, ă, â, î) cannot be encoded in charmap, causing
+    # UnicodeEncodeError when print() is called with OCR-extracted text.
+    if sys.stdout and hasattr(sys.stdout, 'buffer'):
+        sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8', errors='replace')
+    if sys.stderr and hasattr(sys.stderr, 'buffer'):
+        sys.stderr = io.TextIOWrapper(sys.stderr.buffer, encoding='utf-8', errors='replace')
+
     global _paddle_engine, _tesseract_engine, _doctr_engine, _worker_initialized
 
     if _worker_initialized: