Context Navigation

#451 new defect

Method used to set root is very naive (see highlighted code ):

it is set to those 2 target-subtrees that match son-of-root-subtrees of source-. order is random. works ok if trees are similar. doesn't work at all if trees differ significantly.
if group insertion could handle keeled groups (resp. keeled target subtrees), it would not be necessary to change root at all. Unwanted effects (by not doing so):
- if both sons of source-root are a group
  ⇒ results differ depending on target root position: sometimes one group is skipped, sometimes both get inserted
- if one son of source-root is a group
  ⇒ group is (sometimes) placed at wrong son-of-root in destination

root should be corrected using method implemented with #449 (only perform optionally)
insert keeled group if it fits better (see arb documentation for meaning of keeled groups)
- if both sons of root are groups, they should match the correct subtrees (keeled or not) since implementation of #780. When function can handle keeled groups correctly
  ⇒ check problem again
  e.g. happens with LTPs132_SSU.arb when root is changed:
```
Group 'Bacteria' skipped (got superseded by group 'Archaea'; same best target positions)
```

when corrected update [11432]
related: #512
requires: #449