Server restart inside of test.sh does not work as expected

[linusha]

There are several problems with the way we handle server restarts in the test.sh script:

• In the case that the CI env variable is not set, the server is not correctly shutdown.
• In the case that the CI env variable is set, the server is not correctly shutdown when running the script locally. It is unclear why that works remotely, as starting a server locally after running the script does not succeed due to the port already being used.
• I could reliably kill everything by invoking pkill -9 -f '.*lively.*' inside of the same shell that the script ran it previously. However, using the same line inside of the script did not always lead to the same result. I expect that this has something to do with subshells and the UNIX process hierarchy.
• There is however an additional catch: The above command will not work as expected for the case that one starts the a new test server (running test.sh without the CI env variable set while another lively server is running on a port other than 9011), as this kills all lively servers. However, we need a command that only kills the one on 9011 and -n does not cut it, as we need to kill multiple processes.