feat: send SIGTERM to old containers immediately on deploy

The docker-local scheduler now sends `SIGTERM` to old containers immediately after a successful deploy via `docker container kill --signal=SIGTERM`, rather than waiting `wait-to-retire` seconds before signaling. This matches Heroku's graceful-shutdown contract and lets applications begin draining in-flight work as soon as proxy traffic switches. The existing `wait-to-retire` grace period and `stop-timeout-seconds` hard-stop continue to apply unchanged as the authoritative cleanup path.

plugins/scheduler-docker-local/scheduler-deploy

@@ -84,6 +84,19 @@ trigger-scheduler-docker-local-scheduler-deploy() {
 
   # kill the old container
   if [[ -n "$oldids" ]]; then
+    # Send SIGTERM to old containers immediately so the application can begin
+    # graceful shutdown the moment proxy traffic switches to the new containers.
+    # We use `container kill --signal=SIGTERM` rather than `container stop`
+    # because it returns immediately and docker does not follow up with a
+    # SIGKILL on its own when the signal is something other than SIGKILL -
+    # the existing background block below remains responsible for the
+    # eventual hard stop after the wait-to-retire grace period.
+    for oldid in $oldids; do
+      if "$DOCKER_BIN" container inspect "$oldid" &>/dev/null; then
+        "$DOCKER_BIN" container kill --signal=SIGTERM "$oldid" &>/dev/null || true
+      fi
+    done
+
     # Let the old container finish processing requests, before terminating it
     dokku_log_info1 "Shutting down old containers in $DOKKU_WAIT_TO_RETIRE seconds"
     (