ServerRunMode.FORKED always leaks running server processes when cluster shutdown fails

See WFLY-13897 for the 'reproducer' in WildFly.

This is due to a couple of things:

1. Obtaining PID in JDK8 is not implemented.
2. Obtaining PID is broken in higher JDKs due to ISPN-12366.
3. The obtained PID is wrong anyway, because its the PID of the parent shell process, not the child server process.
4. The kill command usage is incorrect and fails anyway.
5. Even if killed, the monitoring process leaks anyway due to ISPN-12367.
6. The stop command doesn't check for client and server errors on the REST call.

The fix is to assign random UUID to a server process and filter properly; use process API correctly.