Track test coverage

As we're starting to stabilize Tide's design, we'll want to work to improve *resilience*. A big part of that is to write tests to prevent regressions. To help inform us which parts of Tide are currently undertested it'd be good if we could have a (documented) way gather test coverage.

Perhaps we may want to integrate it as part of CI eventually as well, but for now simply having a way to measure code coverage would be great.