[CT-83] No test that the container shuts down cleanly under Docker's SIGTERM/timeout behavior

[mftee]

Overview

contract/tests/graceful_shutdown.rs covers shutdown at the process level (see CT-67), but Docker sends SIGTERM and then SIGKILLs after a grace period (default 10s) if the process hasn't exited — worth confirming the service's shutdown sequence (draining in-flight requests, closing the Redis/Horizon connections) reliably completes within that window under realistic load.

Tasks

• Measure and document the service's typical shutdown duration under load in docs/contract-deployment.md
• If it risks exceeding Docker's default grace period, document the required stop_grace_period for docker-compose/orchestration configs

Acceptance Criteria

• Shutdown timing under load is measured and documented, with deployment guidance if it's close to typical orchestrator timeouts

Notes for Contributors

Comment below to be assigned.