Tree modifications have uncontrolled impact on groups

[westram]

Moving subtrees

For normal groups the impact is small in most cases:

• moving one of the 2 childs at the top-level of a group
  ⇒ does also move group information

if a majority of the group is moved	→ ok
if only a minority of the group is moved	→ not ok

For keeled groups (see #735) the impact depends on the orientation of the group:

• if the group was keeled to (old) brother of moved subtree
  ⇒ afterwards it's either keeled
  • to (new) brother or
  • to moved subtree itself,
  i.e. completely wrong anyway
• if the group was keeled to the moved subtree
  ⇒ no impact

NNI

• basic operation swap_assymetric() causes similar problems to those described for Moving subtrees above (did not inspect details yet)