Binning function on pure awkward arrays.

[nilsnevertree]

Very spares arrays are created by the stationary setup of the eurec4a 1d rainshaft example. To handle these datasets, it would be convenient to perform the binning process with pure awkward array methods.

In the long run, a flexible methodology to bin by e.g. superdroplet id, gridbox, time in a flexible order. This would allow to create eulerian and lagrangian calculation e.g.

• Calculate the mass change over time for each superdroplet. For this, it would be good to have the indexers in the following order: superdroplet id, `time``.
• Calculate the mass change in each bin over time. For this the indexer order would be time, gridbox. Along the last axis of variable length, the attributes e.g. mass, radius, multiplicity and also super droplet id would be stored. Summing over the mass along the last axis is convenient on awkward array.

[nilsnevertree]

An idea would be to extent the SupersData Class in a form that attributes and indexers are stored within it. This would require some subclasses which can be then used to wrap around the binning functinos and other handy funcitons as summation or differentiation as in xarray.