Validate newsubjdata argument for posterior_predict, log_lik, etc.

Doesn't check if the correct variables are included (that's done in pp_data), just that newdata is either NULL or a data frame with no missing values. Also drops any unused dimensions in variables (e.g. a one column matrix inside a data frame is converted to a vector).

validate_newdata(x)

Arguments

x	User's 'newdata' argument

Value

Either NULL or a data frame