The way we sample occupancy type is to actually update the uncertainty on all damage functions in the entire damage function family.
At the time of sampling though we actually know the hazard, we should only sample the damage function for the given hazard.
this would require a change to the sample interface to allow for an argument of hazard, not a big deal, but it makes it where it does not conform to the value sampler interface - which is probably ok.
https://github.com/USACE/go-consequences/blob/64ddb7aac89f8cd85a2970e6612be90a9af5b092/structures/structure.go#L52
The way we sample occupancy type is to actually update the uncertainty on all damage functions in the entire damage function family. At the time of sampling though we actually know the hazard, we should only sample the damage function for the given hazard.