environment: NM pid + restart NM + crash detection

unify calling of nm_pid(), no need to be in a try block

create fuction restart_NM_service(), which is better if we decide to track NM PID

rewrite crash detection, detect crashes in every after_scenraio (not only if PID changes), and if there is unreported crash, report it and fail current scenario. Now, FAF and COREDUMP appear before NM log in HTML report.