There are tests of the internal calculation logic, but there are not tests that assert "given this JSON in I expect this JSON out"
It is important to have this coverage for the scenario endpoint, in particular, which does not add any new calculation functionality, but has non-trivial logic to orchestrate and aggregate individual tree calculations.