SIGTERM is not handled: docker stop kills the node without a graceful shutdown #784

Summary

The node only handles SIGINT. SIGTERM, which docker stop (and systemd, Kubernetes) send, has no handler, so the process is killed immediately with no graceful shutdown: no Shutdown signal received, actors and servers are not stopped, nothing is flushed.

Evidence

• bin/ethlambda/src/main.rs:988 (and 994, 998, 1002, 1131) only await tokio::signal::ctrl_c().
• Mainnet beacon follower, main at c3f5f56, after warm-up: SIGTERM ×3 → exit code 143 in ~0.1 s, no shutdown log lines. SIGINT ×4 → exit 0 in 52 ms - 1.27 s.
• Lean devnet node from genesis: SIGTERM → exit 143 as well.
• The README works around it with --stop-signal SIGINT.

Suggested fix

Treat SIGTERM like SIGINT (e.g. tokio::signal::unix::signal(SignalKind::terminate()) selected together with ctrl_c()). The spawned v0.7 adoption PR #783 does this as part of a larger change (SIGTERM shuts down cleanly in 93-102 ms on mainnet), but a small standalone fix could land first.