I particularly like your++ short functional solution.

As for the IE problem… it seems it is more averse to the remove of the original node (also implicite in the original solution with insertBefore.)

So here is a workaround for the IE, avoiding the remove by inserting a cloneNode and display:none the original node.

<script type="text/javascript">
( function() {
    var td = document.getElementById( 'nodelet_container' ).parentNode
+;
    var tr = td.parentNode;
    tr.insertBefore( td.cloneNode(true), tr.firstChild );
    td.style.display= "none";
} )();
</script>
[download]

This being a workaround has a shortcoming: The script interferes (aborts) the cloneNode, effectively cutting the nodelet tower under "Free Nodelet" at its base. So the "Free Nodelet" has to be the last (highest number) in your nodelet settings.

Mixed users (FF/IE/...) might use this out of the box, or can use a browser switch as in:

<script type="text/javascript">
( function() {
    var td = document.getElementById( 'nodelet_container' ).parentNode
+;
    var tr = td.parentNode;
    if (document.all) { // IE
        tr.insertBefore( td.cloneNode(true), tr.firstChild );
        td.style.display= "none";
    } else {  // w3c
        tr.insertBefore( td, tr.firstChild );
    }
} )();
</script>
[download]

also avoiding the free nodelet cut in non-IE.

Hmm, I tried both under IE6 and I only see the XP nodelet and the top of the freenodelet (i.e. the title cell is fine but body cell is empty and compressed vertically). I must say I have new respect for those of you dealing with browser differences!!

I tried both under IE6

The if (document.all) clause in #2 makes sure IE will run the same commands in both. So no surprise they behave identical.

I only see the XP nodelet and the top of the freenodelet

See my remark in the middle of the post ("shortcoming"). Freenodelet has to be your bottom nodelet to not loose other nodelets below it.