The current cleanup code contains the following which Mark has observed has an unnecessary var element = document.getElementById(id). This has not been removed, however, since it is possible that idCallbackTuple.handler assumes the existence of the element:
elementUpdateListeners = reject(elementUpdateListeners, function(idCallbackTuple) {
var id = idCallbackTuple.identifier;
var element = document.getElementById(id);
//test if inner element still exists, sometimes client side code can remove DOM fragments
if (element) {
var updated = (-1 !=
idCallbackTuple.ancestors.indexOf(" " + updatedElementId + " "));
if (updated) {
var callback = idCallbackTuple.handler;

Assign to Mircea to apply the noted improvement.

var id = idCallbackTuple.identifier;
var element = lookupElementById(id);
if (element) {
    var updated = contains(idCallbackTuple.ancestors, updatedElementId);
    if (updated) {
        var callback = idCallbackTuple.handler;
        try {
           callback(element);

If the ancestors contains check is faster than lookupElementById, then we should swap the order. And if we do get the element, we should pass that into the callback, so that the callback functions won't have to immediately find the element themselves for their cleanup.

Reversed test order to make findAndNotifyUpdatedElement function more efficient.