Depend on `primarycensoreddist` in custom family `brms` functions

[athowes]

The initial version of primarycensoreddist was just made available. It is an R package which has the functions:

• dprimarycensoreddist() dpcens(): Compute the primary event censored PMF for delays
• pprimarycensoreddist() ppcens(): Compute the primary event censored CDF for delays
• rprimarycensoreddist() rpcens(): Generate random samples from a primary event censored distribution

We should be able to use these functions to replace our implementations in posterior_predict_family and log_lik_family (I don't think for posterior_epred_family). This would reduce code duplication (we are just replacing the distribution basically in some of these functions, while copy pasting code currently).

[seabbs]

Agree. FYi the aim is to release primarycensoreddist to CRAN by the end of next week. We should also be able to use the stan side code for a marginalised model (and the count based version of that model as well probably more interestingly