Harness

Add failure mode support for:

1. Aborting an entire test set (this will require a feedback mechanism to cause threads to exit - probably best done in the Driver.run() loop; however, if certain threads did not finish in a reasonable amount of time we might have to force them to exit... otherwise the whole thing could hang/deadlock).

2. Run a recovery script (similar to a backout script, but the idea is to execute steps to enable the failing test to be re-run) and then retry (a configurable number of times) with an optional delay (configurable in milliseconds) in between each retry; potential issues to resolve:
- figure out how to represent both the recovery steps and the retries in the report output; it is probably best to implement this as additional steps added to the steps list of the original test (this is how it was done for backout processing)
- figure out how to reset the test state so that a retry could be successful