I am thinking about three separate approaches which should be compared and contrasted:
- Each distribution class implements the hazard rate and cumulative failure function.
- Implement a "Failure" distribution that accesses the necessary components when need.
- Both 1 and 2.