improvements to ksp algorithm

[robfitzgerald]

while running a recent test i noticed two flaws in the logic:

• while it does test for similarity via the similarity function, in the case that similarity threshold = 100.00%, we could accept routes which are in fact exactly the same. in that case, it should still reject it, since we want k unique paths
• the user should be able to set an optional ksp search termination criteria with a reasonable default which allows us to continue searching even when we have found k solutions
• we should have an implementation of Yen's/Eppstein's KSP Algorithm for smaller maps, as the Single-Via approach does not give as many results with smaller maps