Add function to synchronize roots of two trees

[westram]

Motivation:

• ARB_NTREE/Tree/Move groups does a poor job, if the root of the destination is located at a different branch (even if the 2 topologies are identical).
• using similar roots simplifies topology comparison (#444)

Implementation:

Modes:

1. set root of target to an edge similar to root-edge in source
2. set root nodes of two (or more) trees such that
   • root is nearby the optimal root (as provided by ​reroot-mode)
   • both root-edges are similar
     

(Note: 1. will not modify the source ; 2. may modify all trees)

Similarity of 2 root-edges:

• each edge divides the X into 2 partitions (pX₁ and pX₂)
• comparing two edges in two trees (A and B) needs to calculate 4 pairwise differences pA_x Δ pB_y between the 4 partitions defined by the two edges. Two combinations of pairwise differences are possible:
  • pA₁ Δ pB₁ and pA₂ Δ pB₂
  • pA₁ Δ pB₂ and pA₂ Δ pB₁
    
  Only the better combination should be considered!
• comparing 3 edges in 3 trees needs to calculate 12 pairwise differences.
• the value of a pairwise difference might be e.g.

  pA_x Δ pB_y = |pA_x ∪ pB_y| - |pA_x ∩ pB_y|

  where |set| is the size of the set.
• the value of a combination might be
  • the sum of pairwise differences,
  • the minimum pairwise difference (wont work, see comment13)
    or
  • some mix of both.

---

Required by: #734, #451
Related: #512

[westram]

Why the minimum pairwise difference won't work:

Given 2 trees A and B and one species S present in both trees. The leaf edges between S and the rests of both trees define the following partitions:

	pA1	pA2	pB1	pB2
containing	S	rest of A	S	rest of B

Here the minimum would always be

pA₁ Δ pB₁ = 0

⇒ all leaf edges (at species present in both trees) would match perfectly - regardless how the rest of the matches
⇒ leaf edges would be preferred over other edges (esp. for pairs of trees where topology differs).

[westram]

previous change was stupid → reverted.

[westram]

by [18634]