masking of cells in forecast file

[wsavran]

current implementation masks an entire spatial cell and doesn't provide granular access to apply binning for individual space-magnitude cells. the csep1 forecast file allows individual space-magnitude cells to be masked, therefore this should be included

[wsavran]

this can be potentially done by using a numpy masked_array to replace the data currently stored in the forecast. the mask can be populated on initialization of the forecast object, but should be stored/accessed from the region. this way other objects that use regions (eg., catalogs) can take advantage of this masking.

1. mask is defined on the region data structure and used to query from data structure
2. forecast uses mask to sum rates and other data manipulation tasks