Fix both update_tree_view() and traverseUpdate_tree()
A new private var is added for kmk - tree_needs_update - a bool,
which should indicate that treeListView needs update after an "update".
update_tree_view() was depending on specific data ordering - this fixes
that, and now there is only one corner case where it may complain: if any
of the TREE_BASEs are last in the file, there could be missing folders in
the tree view. Even then, there is a qWarning() issued, advising a re-scan.
traverseUpdate_tree() is now near a working state, but maybe it should
try harder to imitate its parent - traverse_tree() - by remembering which
is the last added item... This should happen in a later patch.