fix: apply reviewer feedback to gunicorn config

- preload_app = True: saves memory via copy-on-write in container deployments
- max_requests = 1000 / max_requests_jitter = 50: recycles workers
  periodically to guard against memory leaks, jitter prevents all workers
  restarting simultaneously
- graceful_timeout = 30: explicit clean shutdown window during ECS deploys
- accesslog/errorlog = '-': routes all logs to stdout/stderr for CloudWatch
- worker_tmp_dir = '/dev/shm': uses in-memory fs for worker heartbeat files

Author: Boanerges1996

gunicorn.conf.py

@@ -5,4 +5,15 @@
 threads = 2
 worker_class = 'gthread'
 timeout = 30
+graceful_timeout = 30
 bind = '0.0.0.0:8081'
+
+preload_app = True
+
+max_requests = 1000
+max_requests_jitter = 50
+
+accesslog = '-'
+errorlog = '-'
+
+worker_tmp_dir = '/dev/shm'