fix: flash_backup validation, smoke test assertions, docker/notification test coverage

- storage.py: validate initiateFlashBackup response before returning success=True
- test-tools.sh: remove set -e/inherit_errexit; add success assertion to smoke tests
- test_destructive_guards.py: add confirm-guard tests for new docker destructive actions
- test_docker.py: assert mutation variables in organizer tests; add items branch test
- test_query_validation.py: add 5 missing notification mutation schema test methods
- test_notifications.py: use lowercase importance to test uppercasing logic

Resolves review threads PRRT_kwDOO6Hdxs50FgPb PRRT_kwDOO6Hdxs50FgO4 PRRT_kwDOO6Hdxs50FgO8 PRRT_kwDOO6Hdxs50FgPI PRRT_kwDOO6Hdxs50FgPL PRRT_kwDOO6Hdxs50FgPm PRRT_kwDOO6Hdxs50E2iK PRRT_kwDOO6Hdxs50E2im

scripts/test-tools.sh

@@ -20,8 +20,7 @@
 #   2 — prerequisite check failed (mcporter, uv, server startup)
 # =============================================================================
 
-set -Eeuo pipefail
-shopt -s inherit_errexit
+set -uo pipefail
 
 # ---------------------------------------------------------------------------
 # Constants
@@ -178,6 +177,29 @@ smoke_test_server() {
     return 2
   fi
 
+  # Assert the response contains a valid tool response field, not a bare JSON error.
+  # unraid_health action=check always returns {"status": ...} on success.
+  local key_check
+  key_check="$(
+    printf '%s' "${output}" | python3 -c "
+import sys, json
+try:
+    d = json.load(sys.stdin)
+    if 'status' in d or 'success' in d or 'error' in d:
+        print('ok')
+    else:
+        print('missing: no status/success/error key in response')
+except Exception as e:
+    print('parse_error: ' + str(e))
+" 2>/dev/null
+  )" || key_check="parse_error"
+
+  if [[ "${key_check}" != "ok" ]]; then
+    log_error "Smoke test: unexpected response shape — ${key_check}"
+    printf '%s\n' "${output}" >&2
+    return 2
+  fi
+
   log_info "Server started successfully (health response received)."
   return 0
 }