Snakemake workflow

[thomas-fred]

An attempt to wrap the existing model code in a light-touch snakemake workflow. Should make it easier to reproduce analysis results in the future.

Factoring out library code into local package and trying to clean up formatting, imports, etc. as I go.

Rules written:

• Basic damages
• Calculation of inputs to transport flow allocation (preparation of OD, etc.)
• Nominal flow allocation

To-do notes:

• transport_failure_flow_allocation is the exhaustive transport rerouting rule. It's a thin wrapper of the transport_failure_scenario_setup.py script and invokes multiple instances of transport_failure_analysis.py which currently consumes on the order 70GB RAM. Each network flow allocation (using igraph_scenario_edge_failures) takes a good few seconds. Need to parallelise otherwise network with ~108k edges will take several weeks to run (or otherwise reduce set of edges to fail).
• Water model only has single script, but there's lots more on the cluster in mistral/jamaica-ccri/drought. Don't know where to start there.
• Electricity model core to be installed (see JEM), then flow analysis to be captured in new rule based on Nadia and Tom's work.
• Calculation of EAD and EAEL script requires all the losses from the various network models to be present -- this to be done last.