set of changes related to new heldout-test feature (#300)

[pdreiter]

• tested with and without heldout-test content
• IMPACT: Changes the TestOutcome, s.t. these two cases: (without heldout tests) and (failing heldout tests) are indiscernible

[pdreiter]

This is relevant to butrs-red-team-evaluation/issues/54

[pdreiter]

Based on conversation with @ChrisTimperley - an alternative strategy would be to use darjeeling as an interface to the docker container to evaluate patches.

[pdreiter]

I'll be retooling this approach to reduce the impact to existing darjeeling.