Create a CensoredNormal distribution

[sbidari]

The observation process for the log viral genome concentration in wastewater follow a censored normal distribution with concentration below the level of detection (LOD) censored.
We can use @dylanhmorris's implementation of the censored normal distribution here.

I am wondering if the censored normal distribution in PyRenew should inherit from the RandomVariable class instead of numpyro.distributions.Distribution so that it can be "time-aware" like all other randomvariables. Thoughts? @damonbayer @dylanhmorris @AFg6K7h4fhy2

[dylanhmorris]

I would advocate for a full distribution and then using DistributionalVariable when we want to treat it as an RV