Add automated fault tolerance testing to `run_stress_tests.sh`

[robertnishihara]

We should extend the jenkins tests to simulate node failures (e.g., simulate multiple nodes using multiple Docker containers, and then kill some of the processes in some of the Docker containers).

[robertnishihara]

Some questions:

• Should we give test scripts control over which docker containers are killed and when they are killed or should we just kill containers randomly? If we do give the test scripts control, what's the best way for them to specify when/which containers to kill.
• Should we kill the entire docker containers using docker kill? Or should we kill individual processes within the containers?

[robertnishihara]

We're doing this in the long running stress tests.