Change 'mtr_kill_process' to only take three arguments. The last indicating max number of seconds to wait for process to be killed.
Leave it to the implementation of 'mtr_kill_process' how long to wait between each attempt to kill/detect if process has been killed.