feat(backup): write database dumps gzip-compressed

Dumps stream through gzip into <db>.backup.sql.gz; no uncompressed copy touches the disk. Restore reads gzip and plain dumps by their magic bytes, verifies a gzip dump to its end before anything is dropped, and still finds the uncompressed names of older generations. The manifest names the compression.

🎯 Fix: 92 %
💥 Break: 20 %
💸 Cost: –
⏱️ Time: –
🛡️ Security: ~100 %
🏆 Score: 74
🚦 Gate: ⛔ break 20 % > 5

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/e2e/test_e2e_mariadb_anonymous_preemption.py

@@ -34,6 +34,7 @@ from .helpers import (
     MARIADB_DATA_DIR,
     MARIADB_IMAGE,
     cleanup_docker,
+    dump_text,
     require_docker,
     run,
     unique,
@@ -159,8 +160,7 @@ class TestE2EMariaDBAnonymousPreemption(unittest.TestCase):
             self.assertTrue(produced, "backup_database did not produce a dump")
             dump_path = Path(volume_dir) / SQL_DIR / f"{self.db_name}{DUMP_SUFFIX}"
             self.assertTrue(dump_path.is_file(), f"expected dump at {dump_path}")
-            with dump_path.open(encoding="utf-8", errors="replace") as f:
-                content = f.read()
+            content = dump_text(dump_path)
             self.assertIn("INSERT INTO", content)
             self.assertIn("'ok'", content)