New splits for profmark

[spetti]

This version of create-profmark can call the newly implemented blue (multi-round random election process), cobalt (greedy algorithm with random order), and independent selection (random selection of test set) algorithms for splitting sequence data into training and test sets. There is also a new script for benchmarking DIAMOND.

This will only work with the version of easel that I will submit in a pull request shortly.

[npcarter]

This passes all functionality tests, so I'm going to merge it into develop. There is a memory leak problem in create-profmark.c, which I will fix in the develop branch.