Improve lab shutdown speed

[maxonthegit]

At present, it takes a while to stop a running lab by using lcrash. Indeed, the script makes vcrash wait to check that every virtual machine has properly shutdown, consuming typically more than 1 second per virtual machine.
This shutdown time could be improved, for example by immediately attempting to kill the machines or by just skipping this check (even if I would be happier to avoid the latter solution).

The same shutdown time improvement may be applied to single virtual machines: provided that the process is stable enough, kill commands may be removed from vcrash and be retained in vclean only.