There are 2465 tests.
This is a bit much for manual verification.
I am doing the same thing that I have done in the past.
Which is to just generate the expected results based on the current output.
The problem is that with this design the tests will pass even if there are problems or mistakes.
These tests will have to be verified and fixed over time as needed.
Porting this overr to the 0.7.x development branch will help me catch such problems.