This is the first change to the Cat's logic for a while, and I'm fairly nervous because the test coverage here is not that broad (essentially we are chasing corner cases) and I haven't looked at the code in ages. @pkoppenb could you have a look and see if the output makes sense to you using a couple of other analysis use cases? I.e. that in situations where there was no problem and this MR should therefore not change anything, indeed nothing changes? If you want to put some options and input files on eos I will happily make some tests from them for the future.
Vava Gligorov authored