Implementation of path scoring based more/less probable edges count

[domainexpert]

… better probability in each branch point

• The path selection is to be used for sensitivity analysis
• The chosen path will have test<N>.choice file generated for it

@Himeshi I think this implementation implements the selection criteria that we want: We select the path that always has better execution probability at every branch point.

[domainexpert]

@Himeshi So the latest commit modifies this PR to instead produce *.improb file for each path containing the count of edges with less probability taken. I hope this will be more useful, and thanks for the feedback!

[domainexpert]

@Himeshi This approach not to be used.