Sampling uncertainty for under ascertainment

Currently we used a fixed baseline CFR for the [ascertainment calculation](https://epiverse-trace.github.io/cfr/articles/estimate_ascertainment.html), but it would be more accurate to instead sample from an underlying estimate (e.g. using {epiparameter} to extract a normal distribution from published mean/95% CI) so the method propagates both uncertainty in currently observed and previously reported CFR.

Start with the linked ascertainment calculation article and trace where the fixed baseline CFR enters the method. Review how epiparameter could extract a normal distribution from a published mean and 95% CI; done means uncertainty from both currently observed and previously reported CFR is propagated through the ascertainment calculation.

Written by the indexing model from the issue text.