[r] hotfix for release directory JSON parsing problem (#649)

Our way of converting the release directory JSON to a convenient data frame was too brittle for recent updates to release_directory.json. Fix by switching from `rbind` to the more flexible `dplyr::bind_rows`. (Our package already transitively depended on dplyr via Seurat and others.)

api/r/cellxgene.census/DESCRIPTION

@@ -16,6 +16,7 @@ RoxygenNote: 7.2.3
 Additional_repositories: https://tiledb-inc.r-universe.dev
 Imports:
   aws.s3,
+  dplyr,
   httr,
   jsonlite,
   methods,

api/r/cellxgene.census/NAMESPACE

@@ -9,6 +9,7 @@ export(get_source_h5ad_uri)
 export(new_SOMATileDBContext_for_census)
 export(open_soma)
 importFrom(aws.s3,save_object)
+importFrom(dplyr,bind_rows)
 importFrom(httr,build_url)
 importFrom(httr,parse_url)
 importFrom(jsonlite,fromJSON)

api/r/cellxgene.census/R/release_directory.R

@@ -28,6 +28,7 @@ get_census_version_description <- function(census_version) {
 #' @return Data frame of available cell census releases, including location and
 #'   metadata.
 #' @importFrom jsonlite fromJSON
+#' @importFrom dplyr bind_rows
 #' @export
 #'
 #' @examples
@@ -50,7 +51,7 @@ get_census_version_directory <- function() {
   raw <- simple_rapply(raw, function(x) ifelse(is.null(x), "", x))
 
   # Convert nested list to data frame
-  df <- do.call(rbind, lapply(raw, as.data.frame))
+  df <- do.call(dplyr::bind_rows, lapply(raw, as.data.frame))
   rownames(df) <- names(raw)
   return(df)
 }