fixed distance clustering algorithm

[corybrunson]

Singh &al emphasize that the number of clusters should not be fixed, but also that domain-specific clustering methods may be used. Currently two intuitive rules are available, based on continuous and discrete versions of the same principle. While it might not usually be appropriate in practice, having a method available that simply imposes a fixed-distance cutoff, i.e. performs clustering without adjusting for the size and scale of each level set, could be useful for demonstration and comparison purposes.

I'm thinking about implementing this, but it's not a priority, and i wanted to check first!