support predictive test selection with test runner

What is the problem this feature will solve?

Very large projects that have a lot of tests take a very long time to run the full tests suite.
Running all of the tests can be avoided, by doing Predictive Test Selection to select only a subset of tests to run.
This enables speeding up the Continuous Integration turnaround time for users using Test Runner.

What is the feature you are proposing to solve the problem?

Some interface, whether programmatic API or CLI, to do predictive test selection.
If we look at Jest as an example (of a popular testing framework), they have:
--onlyChanged and --changedSince, which analyze the (transitive) source code dependencies of the tests and of the changes in Git/Hg, to select only tests that have a source dependency on the changes.

What alternatives have you considered?

Not sure we'll want to lock in users to Git/Hg, some may want to use a different version control solution, so perhaps the integration with the specific source control can be left up to the community. What Node needs IMO is some interface for such library author to use.

No implementation files or tests are named. Start by reviewing Node's Test Runner API and CLI entry points, then compare the proposed interface with Jest's --onlyChanged and --changedSince behavior; done means an agreed design for predictive selection and its source-control integration boundary.