The bug is real but it's not exactly well described.
What YifanHU multilevel algorithm does is that it groups nodes in clusters to fasten layouting. As a result new meta-nodes are created during the process. These meta-nodes don't have any attributes and have default size 1 (tiny). When the algorithm finish it ungroups these meta-nodes and therefore meta-nodes are deleted.
With you dataset it appears that Yifan-Hu Multilevel doesn't finish well and leave meta-nodes. You can see that they remains by clicking on "Hierarchy". Instead of a unique Level, you have two levels.
The bug is real but it's not exactly well described.
What YifanHU multilevel algorithm does is that it groups nodes in clusters to fasten layouting. As a result new meta-nodes are created during the process. These meta-nodes don't have any attributes and have default size 1 (tiny). When the algorithm finish it ungroups these meta-nodes and therefore meta-nodes are deleted.
With you dataset it appears that Yifan-Hu Multilevel doesn't finish well and leave meta-nodes. You can see that they remains by clicking on "Hierarchy". Instead of a unique Level, you have two levels.
I'll rename the bug accordingly.