Restart server automatically if crashed during test

The new server admin API would allow easier server restart in case of a crash during the test. This way, the test can continue even if the grammar contains a known crashing query, without having to explicitly filter out the crashing queries one by one.

This can be accomplished using a Validator that checks for the STATUS_SERVER_CRASHED code and then uses the server admin API to restart the server. Ideally, the log files should be preserved in place as they were, so that they can continue growing as they did before. There should be no need to copy the logs or otherwise move files around. The core file on Linux already contains the PID in its file name, so no action should be required there as well.

Printing the backtrace to STDOUT would still be nice -- currently it will only happen via the Backtrace Reporter at the end of the test, on whichever core file is passed to gdb.