default value isn't used by epi_lag or epi_ahead

[dsweber2]

Fairly minor problem but neither actually passes the default value through add_shifted_columns to epi_shift_single

[dsweber2]

see

epipredict/R/step_epi_shift.R

Lines 245 to 250 in de6e1db

	ok <- object$keys
	shifted <- reduce(
	pmap(grid, epi_shift_single, x = new_data, key_cols = ok),
	dplyr::full_join,
	by = ok
	)

soft block by #296, since that refactors these into a single function, which would make this significantly more awkward

[dajmcdon]

The set of arguments is meant to match https://recipes.tidymodels.org/reference/step_lag.html. But there are no values to fill. It is safe to remove this.

[dsweber2]

I was going to add support for it, since that's basically the same amount of work; would you rather it be gone?

[dajmcdon]

Actually, "support" is probably better if it's not too difficult. I think NAs get introduced on the left_join()? And then it's just a matter of replacing them (and not others).