Using Data / Empirical Distributions inputs to get delay distributions

[pearsonca]

Per conversation w/ @seabbs:

I am using EpiNow2 in concert with a stochastic model which particular discretized time distributions for various events. I want to ensure that those time distributions match the delay distribution inputs to EpiNow2, so I essentially sample the stochastic model distribution to generate event times, then put that in dist_fit(...).

Turns out for my use case, I should use estimate_delay or bootstrap_dist_fit.

This indicates there's a bit of an engineering issue with the interface. I recommend not exposing dist_fit. Also, estimate_delay is a bit odd as a wrapper - it does literally nothing more than bootstrap_dist_fit except hard enforce one of the default arguments. I think that ends up being more confusing - if there was some transformation of inputs or outputs, it would make sense to have a wrapper function, but that's not the case.

If you want to preserve the name space for a future wrapper (i.e., some day you see bootstrap_dist_fit as having different arguments), that's fine - but probably makes sense then to only export estimate_delay.