Since the old IDL version grew slowly over a couple of years, I chose not to include all features in the port to Python. If you were using a feature that was available in the IDL version and not in the Python version, please let me know here and I will try and add it back in a future release.
I have some ideas about the ACOLITE outputs, they are only for references and may be not totally correct.
(1) Outputs are the L1R, L2R and L2W, these 3 files may be merged as one file, the L2R always includes L1R (rhot_*) and L2W is just masked surface reflectances and water quality.
(2) File size of output file is too big, in particular for the Sentinel-2, the output is greater than several Gb. Is it possible to use scale factor (for example, 0.0001) and int datatype in the dataset?
(1) There is some overlap between L1R and L2R, but the L1R file is sometimes used in the processing. I could add a delete L1R keyword which removes this file after processing. L2R and L2W do not necessarily have the overlap you describe; for example, the L2W could also just contain a turbidity product.
(2) I could add another keyword to convert the L2W data to integers. At the moment if you don't need the 10 m resolution for Sentinel-2 you can always process at 20 or 60 m.
I'm not sure when these things will arrive, but I'll keep them in mind!