Meta: Refactor code base

[johnlees]

Things to be done off the top of my head:

• Remove dbFuncs (Remove dbFuncs dictionary #166)
• Remove all use_mash instances, and the partial functions
• Redo command line options (Switch to docopt #73)
• Move visualisations out (Separate '--generate-viz' mode into an exported script. #77)
• Integrate lineage assignment with sketchlib functions and model objects (Make lineage assignment into a model object #79)
• Rewrite the documentation, probably from scratch (Update documentation #74)
• Make a simpler assign_query API
• Write better tests
• Move assign_query out (Make assign query its own program #107)
• Make network into a class (see Combine all the network functions into a class pp-netlib#2)
• GPU use a single option rather than three (all, and catches error if sketching assemblies)

[nickjcroucher]

Happy to take on removal of mash/partial functions, moving visualisation code and test updates - they seem to need your expertise least?

[johnlees]

I became increasingly aware with the latest changes that a lot of parts of the code now are not as we would write them if doing this with our current tools. I think the models are broadly ok, but the sketching, distances, networks and visualisations have all changed a lot since the first version. I think all of __main__ could basically do with a comb through at some point, although it doesn't have to be right now.

I'm happy with whatever you'd like to do!

• Removing use_mash/partials would be a prerequisite for the first point.
• With the tests at least checking the cluster results are the same each time would be a start.
• I should probably do Switch to docopt #73, Update documentation #74 and Make lineage assignment into a model object #79, at least.