Functionality to support users with modelr::data_grid into newdata

[athowes]

The workflow (see #52) we have for creating newdata (that goes into tidybayes) is:

newdata_sex <- obs_prep |>
  as.data.frame() |>
  modelr::data_grid(sex) |>
  mutate(obs_t = NA, pwindow_upr = NA, swindow_upr = NA)

Perhaps we want to support users to do this somehow?

Maybe just the mutation part? It could have options to specify e.g. "no censoring" or "no truncation" rather than having to know what to do yourself? (But also we should make it more obvious via the names.)

[seabbs]

I think ideally it would dispatch on the class of the model people have used

[seabbs]

I also would be surprised if that as.data.frame was actually needed?

[athowes]

I also would be surprised if that as.data.frame was actually needed?

I think it's needed currently but I don't think that it should be needed.

[athowes]

I think ideally it would dispatch on the class of the model people have used

Yes I agree. Given that the parts that are required are the LHS of the formula which are model specific. It could be that in future we allow users to "predict without censoring" using a different model in a different way than by setting window variables to zero.

[athowes]

Flagging that I'm using these code snippets, and I think would be good to have this as feature in the package (with dispatch on model class).