fix(procmgr): terminate residual job members before deferred drain

When the root child exits while descendants remain in the supervision job,
call TerminateJobObject before retaining the job handle for async profile
unload so KILL_ON_JOB_CLOSE is not the only way to reap lingering members.

Author: jose-manuel-almaza

pkg/procmgr/rust/src/process.rs

@@ -239,6 +239,12 @@ impl ManagedProcess {
                 }
                 Ok(_) => {
                     if let (Some(pid), Some(profile)) = (exited_pid, self.user_profile.take()) {
+                        if let Err(e) = job.terminate() {
+                            warn!(
+                                "[{}] failed to terminate residual job members before deferred drain (pid {}): {e:#}",
+                                self.name, pid
+                            );
+                        }
                         let job = self.job_object.take();
                         info!(
                             "[{}] deferring profile unload until job members exit (pid {})",