Fix admin healthcheck: use 127.0.0.1 instead of localhost

Alpine Linux resolves localhost to IPv6 ::1, but Next.js binds to
IPv4 0.0.0.0 — causing the healthcheck to fail with connection refused.
Also update worker env vars from legacy Celery names to current ones.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docker-compose.yml

@@ -118,7 +118,7 @@ services:
       api:
         condition: service_healthy
     healthcheck:
-      test: ["CMD", "wget", "--no-verbose", "--tries=1", "--spider", "http://localhost:3000/admin/"]
+      test: ["CMD", "wget", "--no-verbose", "--tries=1", "--spider", "http://127.0.0.1:3000/admin/"]
       interval: 30s
       timeout: 10s
       retries: 3
@@ -164,9 +164,10 @@ services:
       DEFAULT_FROM_EMAIL: ${DEFAULT_FROM_EMAIL:-Casera <noreply@casera.com>}
       EMAIL_USE_TLS: "${EMAIL_USE_TLS:-true}"
 
-      # Worker settings
-      CELERY_BEAT_REMINDER_HOUR: ${CELERY_BEAT_REMINDER_HOUR:-20}
-      CELERY_BEAT_REMINDER_MINUTE: ${CELERY_BEAT_REMINDER_MINUTE:-0}
+      # Worker settings (UTC hours for scheduled jobs)
+      TASK_REMINDER_HOUR: ${TASK_REMINDER_HOUR:-14}
+      OVERDUE_REMINDER_HOUR: ${OVERDUE_REMINDER_HOUR:-15}
+      DAILY_DIGEST_HOUR: ${DAILY_DIGEST_HOUR:-3}
     volumes:
       - ./push_certs:/certs:ro
     depends_on: