Add validate_as_xmap() helper which wraps internal validate_xmap()

[cynthiahqy]

TODO

• [x] add verify_links_as_xmap()
• [x] add tests on well-formed and poorly formed crossmaps

To simplify error handling,

• make validate_xmap_df() internal function (i.e. swap df_check() for lightweight boolean checks, and if any fail, direct user to use validate_as_xmap()
• implement a validate_as_xmap() generic in the style of as_xmap()

It doesn't really make sense to have xmap_validate() functions since the xmap have to be validated before becoming an xmap; and if you made it some other way, then it still makes sense that you want to validate_as_xmap(). The most likely workflow is:

links |>
  validate_as_xmap()

not

links |>
  as_xmap() |>
  xmap_validate()

It could make sense though to have validation for special types #88 e.g.:

xmap |>
  xmap_validate_as_recode()

with additional types:

• xmap_validate_as_agg()
• xmap_validate_as_disagg()

[cynthiahqy]

Generic: validate_as_xmap()

#' Validation Helper for `xmap` objects
#'
#' @param x A candidate object (i.e. data frame) to be validated
#'
#' @return `x` if validation passes. Throws error otherwise.
#'
#' @section Methods:
#' \Sexpr[stage=render,results=rd]{generics:::methods_rd("xmap_validate")}
#'
#' @export
#'
validate_as_xmap <- function(x) {
  UseMethod("validate_as_xmap")
}

[cynthiahqy]

This would allow the crossmap approach to be integrated into existing data preprocessing pipelines

• e.g. for (#108) you could just check the existing lists, and not change any of the other code
• collapse lists to df of links (#66), then validate

[cynthiahqy]

• [ ] refactor validate_xmap_df() to skip or use lightweight check with redirection to validate_links_as_xmap()
• [ ] refactor as_xmap_df() to use validate_links_as_xmap() and new_xmap_df() --- something to reduce check redundancies?