Store map_overlap dask computation in original numpy array

Question

I want to achieve the following workflow (as an edge case in a larger project):

1. Create Dask Array from 2D numpy array
2. Correlate with map_overlap using