optimize scenario run time [4 days]

[perrygeo]

Currently breaks down as follows (on my dev VM, times very approximate using 1.25 mil iters and 20 runs):

• ~ 1 min saving (the Scenario.run() method)
• ~ 3 minutes writing the input .dat files (marxan.py called from tasks.py)
• ~ 5 minutes on the Marxan Pre-processing section ... related to the number of planning units and conservation features ... what else can we do here?
• ~ 2 minutes on the Marxan processing section ... not much to do other than tweak the MARXAN_NUMREPS and MARXAN_NUMITNS
• ~ 1 minute compiling the results (the Scenario.results method)
• 15 seconds transfer
• 15 seconds render

All of these sections have opportunites for increasing speed and/or caching. Need to profile and attack each of them.

[perrygeo]

As of be82918 , the scenario.run() method is much faster (without variable geography) as we globally cache many of the expensive operations. Cache invalidation is now a bit quicker too. This brings the scenario save time down to ~7 seconds.