Use dask cluster in analyze_ensemble.py

[FariborzDaneshvar-NOAA]

@SorooshMani-NOAA as we discussed before, analyze_ensemble.py might not successfully run for big storms/large ensemble due to memory issue.
A quick fix, as you suggested, was adding dask cluster under the if __name__ == '__main__': and before _analyze(tracks_dir, analyze_dir, mann_coef) as follow:

cluster = SLURMCluster(cores=16,
                       processes=1,
                       memory="500GB",
                       account="nos-surge", # "nos-surge" for Hercules, "compute" for PW 
                       walltime="04:00:00",
                       interface="eth0", # only for PW
                       header_skip=['--mem'])
cluster.scale(6)
client = Client(cluster) 

And then running the script manually.

Could you please add this to the main script, and maybe with user defined inputs in input.conf? thanks

[SorooshMani-NOAA]

sure @FariborzDaneshvar-NOAA. One thing we need to test though, is what happens if we run this Dask code in the singularity container. I'm not sure how that would work with running new instances, etc.