Processing and analyses of ragged arrays

[selipot]

As we write analysis functions, I would like to discuss how we need to deal with ragged array formats? Some thoughts:

• Expect the user to manually loop over the indices using the rowsize variable?
• Convert ragged array variables to xarray objects filled with nans as needed and apply vectorization, allowing chunking? (sounds cumbersome)?
• Convert ragged array variables to awkward arrays?