0.4.0
What's new:
- a dict-like class has been added that elegantly takes care of defining and updating the distributions that are used to marginalize over the diagnose times
- the API for the likelihood got an overhaul, using the mentioned new class. It should be much more readable and easier to use, since one does not have to worry about providing time distributions as arguments anymore
- the API for the risk was changed similarly. Therefore, the most important user-facing methods have gotten a facelift
Unfortunately, I currently do not have the time to update the entire test suite. I managed to write a somewhat realistic integration test, but dropped some unit tests. I will at some point catch up on this.