Performance of optional functions before binning
It turn out the using xarray induces quite an overhead in some cases. In !89 (merged) and !87 (merged) we use xarrays highly optimized grouping algorithms to reduce the data. However, for element-wise array operation there is overhead coming from xarrays data labeling (http://xarray.pydata.org/en/stable/computation.html#wrapping-custom-computation). The solution is to either wrap the functions using xarrays ufunc functionality, or move to numpy alltogether within the indicated block in the tbdet method "process_dssc_data".
In case one of the optional manipulation is selected the load_chunk_data() method would return an numpy.ndarray instead of a xarray.Dataarray. From there on only numpy specific functions would be used. On top of that further reduction of overhead might be possible by avoiding frequent reassignment of variables. At the end of the optional block the resulting np.ndarray needs to be cast back into an xarray.DataSet for the following binning routines.