This adds xarray mainly to see if this is the way we want to handle the metadata. Some questions:
Do we want to suffer the complexity of using these rather than numpy arrays?
Is the dataarray correctly structured?
One can imagine piping the data into xarray more efficently, but I would suggest leaving that exercise for later as it is much easier to address after the coming changes to databroker.
Do not merge.
This adds xarray mainly to see if this is the way we want to handle the metadata. Some questions:
One can imagine piping the data into xarray more efficently, but I would suggest leaving that exercise for later as it is much easier to address after the coming changes to databroker.