Encapsulate dealing with "all missings" into a function

[hfrick]

The pattern to check if all the observations in new_data have missing values and (if necessary) return an NA-filled tibble early should probably be its own function.

Example:

censored/R/proportional_hazards-survival.R

Lines 132 to 137 in 1a891a7

	all_missing <- n_missing == n_total
	if (all_missing) {
	ret <- predict_survival_na(time, interval)
	ret <- tibble(.pred = rep(list(ret), n_missing))
	return(ret)
	}