Function run_with_backend_timeout 

pub fn run_with_backend_timeout(
    command: Command,
    config: DetConfig,
    print_summary: bool,
    print_summary_to_json_file: &Option<PathBuf>,
    backend: Backend,
    timeout: Option<Duration>,
) -> Result<ExitStatus, Error>

run_with_backend with a hermit-enforced wall-clock bound on the guest.

On ptrace it needs no particular stack from the calling thread; see run.

This library never installs a process alarm. Ordinary ptrace keeps the original runtime, operation, failed GlobalState and backing guards when a finite cleanup attempt is unconfirmed; downcast HermitCleanupUnconfirmed and recover it on the original process/thread. A timeout is measured before startup and is not a certificate of cleanup. None leaves execution unbounded, but fatal cleanup attempts remain finite. Nonordinary backends retain their existing cooperative timeout/drop semantics. Synchronous stalls require caller-owned supervision; only the CLI owns its hard alarm fallback.

⚠️ NOT PART OF DetConfig, DELIBERATELY. A wall-clock bound is host state, and DetConfig is the determinism configuration that is serialized to disk and reasoned about as guest-visible. Threading a real-time deadline through it would put a nondeterministic quantity inside the structure whose entire job is to exclude them.