Test: reproduce stale content mounting after moving a dynamic group

[nguyenyou]

Recreates the original PR #210, which was automatically closed when my fork was accidentally deleted.

Adds regression tests for NestedGroup.moveToParent.

A child <-- inserter retains its DOM content while unmounted and must reconcile to the latest value on remount. The old node must not receive a mount callback after it has been superseded.

Current behavior mounts the stale node first, then creates and mounts the new node.

Expected behavior is to unmount the old node, create the new node, and mount only the new node.

The stale mount can cause permanent damage

Mounting stale content also runs its own bindings. In the second set of tests, the stale element has a child.maybe <-- that claims an element e living in an unrelated live list:

1. After the move, the stale element mounts and its binding steals e from that list.
2. The group then re-renders and drops the stale element, and e leaves the page with it.

e is never restored. The actual events are [mount:c, unmount:c, unmount:e], where nothing should happen.

The same scenario without a move (the REFERENCE tests) leaves e untouched.

This fails on all three move paths:

• stolen from an unmounted host into a mounted list;
• moved onto a mounted element;
• moved onto an unmounted element that is then mounted.

It also fails at depth 2, where the stale content is a nested group (actual: [unmount:e]). The recursive move re-registers the inner group too, so a fix must hold at every nesting depth, not just for the outermost group.

Run:

sbt 'testOnly *NestedGroupActivationOrderSpec'