Some of the new tests added in #18 use tuple representations which (1) are created manually in tests, and (2) are opaque w.r.t. their semantics.
Is there a reason we cannot do comparisons with Solution objects? Are Solution objects difficult to work with and/or debug?
