Addition of lineage clustering

[nickjcroucher]

Already merged in master so should not require any further alterations.

[johnlees]

Extra request: would you also be able to add in a test for the new mode(s), probably with --use-mash too. No need to check the results, just that they run

[nickjcroucher]

Now using multiprocessing for ranking for all isolates in parallel, and networks for storing relationships, in 905e25b. Everything should now be ready for merging in to master. The networks are not really used yet, and we probably need a rethink of the 'printClusters' function, and how it can be updated and merged with the new 'printLineages' function. This might well be worth doing alongside making the visualisation mode an external script..

[nickjcroucher]

Made some small adjustments to the minimum k-mer length warnings. We should probably mirror these across mash.py and sketchlib.py, and specify the actual minimum values, and put in a message about the use of non-continguous k-mers at the appropriate lengths (wasn't sure of the precise values to get this right in this commit).