Fix PlausibleClusterConnectionStrategy for subduction clusters

[voj]

Part of #300

PlausibleClusterConnectionStrategy has a checkBuildThreaded() pre-calculation step that creates the list of possible jumps to be taken by ruptures.

This step takes 6 minutes for our small fault model sample. Also, for each short enough jump from a crustal cluster to a subduction cluster, this code will potentially check all possible crustal-style permutations on the subduction cluster for Coulomb plausibility. These permutations will not adhere to our rectangular definition in DownDipPermutationStrategy rendering them not just costly but also meaningless.

We want to write our own custom code to create possible jumps to a subduction cluster. The shortest path between clusters is the simplest approach that we can take. We can run plausibility filters on it at a later stage if we're unhappy with the result.

[voj]

Parked until we get more input from scientists on the shape of joint ruptures.