KNOWN BUGS:

- (Small) memory leak in VisWindow.setTitle:
  (actually in VisMain.cpp in VE_SET_WINDOW_TITLE event handling)
- Memory leak in p4vasp.StructureWindow show/hide
- 3D output is not colored correctly when using some versions of mesa
  specular light too strong ?
  Probably badly normalized normals in GLU quadrics.
- Mysterious blue screen of StructureViewer -
  blue backgroun color, and correct cell shape is vissible, but no atoms and no bonds
  Sometimes (mainly for large inputs) ends with Segmentation Fault.
- Hidden atoms and bonds sometimes appear when arrows are shown.
- StructureWindow.win.saveScreenshot() does not work reliably when several StructureWindow-s
  are open.
  
This seems to be working now:
- cStructure.toString - output is wrong when selective dynamics (transformposcar.py)
- Structure.selected can be set, but .toString result shows T T T for all atoms (transformposcar.py)
- VisSync hangs when built with threading, but not running in thread.
- selective is not replicated in p4vasp.Structure.replicateCell
- Segfault on VisWindow destruction - e.g. if StructureWindow is deleted. I was not able
  to identify the problem. The segfault occures on calling e->flwindow->hide() in VisMain
  event handling, when the VisWindow is already deleted, but VisFLWindow still exists.
  It happens always when closing StructureApplet in the frame, but neither in a simple
  script using StructureWindow, nor in the c-program (viewer3).
  Since I need to add some new features, I need to fix this quickly.
  So I fixed it the way, that VisWindow is not destroyed, it is rather appended to
  p4vasp.StructureWindow.zombie list. Yes, very dirty.
