Strahler: Usage of inconsistent branch definition

[tferr]

Several issues exist with current Strahler implementation:

• The detection of branches by inference of n. of end-points is rather fragile. It works for order 1 but, for higher orders, it may or may not be verifiable, depending on the topology of the pruned skeleton
• Analyzing the pruned skeleton is also problematic because we may use mixed definitions of branch. We should ensure that we define branches unequivocally at every iteration. Right now, e.g., root order branches delimited by two junction points are not considered.

For consistency, a branch must always be defined as a segment delimited by either:

• Two end-points
• An end-point and a junction-point
• Two junction points

[tferr]

Should be solved with release release 1.1.4