Model validation/Sensitivity analysis performance metrics

[barneydobson]

Currently need to revisit code to compare a synthetic network vs real one. Then need to determine what metrics to use for sensitivity analysis. A few options and comments:

• Total flooded area - pbias
• Outflow - NSE (though actually matching up with the outflow is not trivial given that we are synthesising these).
• Flow along pipes - NSE (buffer a pipe from real network, match all pipes within the buffer, calculate NSE for each - from a sensitivity analysis perspective maybe use median NSE - this will be an abysmally poor metric I would assume so possibly NSE is not appropriate).

[barneydobson]

Superceded by more detailed issues, summarised by #49