Dummy Sampler which supports no distributions. This is intended for use by models
not using a helper library which therefore need to implement their own Sampler
subclass and may need to only support a small number of distributions (or none),
and thus can use DummySampler as their Sampler implementation or a base for their
own subclass which only overrides methods as needed.
distIsSupported always returns false and all the sample* methods throw an
UnsupportedOperationException.