`numpy.savez` raises TypeError when saving aggregated data

[wcarthur]

Running on raijin, with the following modules:

module load openmpi/1.6.3
module load python/2.7.11
module load python/2.7.11-matplotlib
module load geos
module load gdal/1.11.1-python

Using wind_v5 template on permute branch.

Traceback (most recent call last):
  File "../../core_hazimp/hazimp.py", line 72, in <module>
    start(config_file=CMD_LINE_ARGS.config_file[0])
  File "../../core_hazimp/hazimp.py", line 64, in start
    the_pipeline.run(cont_in)
  File "/short/w85/cxa547/hazimp/core_hazimp/pipeline.py", line 69, in run
    job(context)
  File "/short/w85/cxa547/hazimp/core_hazimp/workflow.py", line 54, in __call__
    self.job_instance(*args, **job_kwargs)
  File "/short/w85/cxa547/hazimp/core_hazimp/jobs/jobs.py", line 708, in __call__
    use_parallel=use_parallel)
  File "/short/w85/cxa547/hazimp/core_hazimp/context.py", line 207, in save_exposure_aggregation
    numpy.savez(filename, **write_dict)
TypeError: savez() keywords must be strings

The numpy.savez documentation indicates the **write_dict should have keys that are strings.