Print profile dump before killing worker when we timeout

[nickrobinson251]

When a testitem hits the timeout limit, we print out whatever logs it produced and then kill the worker, but we don't print out any info about what the worker was actually doing, which makes it hard to know e.g. if the testitem was deadlocked.

It would be helpful to collect and print out a CPU profile by sending a SIGINFO/SIGUSR1 to the process before sending the SIGTERM

Alternatively, we could dump julia task backtraces and/or CPU stacktraces.