XMLTokenizer's wrapped mode used with grouping does not generate reasonable results from most user's point of view.
More concretely speaking, Its wrapped mode generates a sequence of tokens, each token element wrapped in its context elements (i.e., the parent elements). This is appropriate when no grouping is used. However, with grouping used, it results in a token consisting of a group of such context-wrapped tokens. A more intuitive and user-friendly result would be to have a token consisting of the original token elements wrapped in their common context elements. In other words, a token must consist of multiple token elements in their common wrapped context.