This enables us to process text() and @attr results.

* Make the XPath query for the "Remove XML child nodes" operation required.
* Don't dumb down language because it'll hunt us sooner or later.
  Which mostly means renaming XML elements to XML nodes.

I guess this concludes our "modify XML" series for now.

Also works with XPath, like the insertion operation.

The XML data wrapper now tries to be less smart and more consistent,
always returning a DOMNode as data() and not returning "artificial" elements
like tag information and attributes. (Those can be extracted with a new,
separate operation.)

In addition, there is now an XPath operation, which makes XML imports
a lot more feasible than before.