Add the ability for sippy to process incoming test output and on failure, if the test is configured for it, try to extract data from the test output into generic database structure.

This would be a mechanism to find what next needs to be broken out of a generic backstop test. (alerts, pathological events, etc)

We likely need multi-valued, as these tests often fail for multiple reasons within their test output.

We could link prow_job_run_test to multiple rows, prow_job_run_test_dimensions. Or should we use a json column of some kind for generic storage?

Stephen has a good start on this behavior with his queries for extracting strings from test output in sippy db via regex. (link TODO)