Force xcube Server to open multi-level datasets

[forman]

• Changed xcube Server impl. to force opening multi-level datasets.
• Added schema for "data_type" parameter fro all filesystem data stores

Checklist:

• Add unit tests and/or doctests in docstrings
• Add docstrings and API docs for any new/modified user-facing classes and functions
• New/modified features documented in docs/source/*
• Changes documented in CHANGES.md
• GitHub CI passes
• AppVeyor CI passes
• Test coverage remains or increases (target 100%)

[konstntokas]

Looks good. I tried to test it on the xcube server demo, but could not recognize any performance increase. Maybe the demo example is too small.

Also, it will be kind of hard to test that "mldataset" is selected, because the output of the function _open_ml_dataset stays the same. Would be interested in how you'd do it.

[konstntokas]

As far as I understand this is the case for datasets calculated on the fly (in memory). Do we need to take care, that this is done as mldatasets as well, or is this covered already?