Util for programmatically creating dataset for processed files

[anngvu]

Is your feature request related to a problem? Please describe.
Requirement is to organize processed products as datasets, but the experience of adding files from the UI isn't that great especially when the output is spread across different folders/several folders deep.

Describe the solution you'd like
Something like new_dataset that takes parameters 1) an output location (from a standard workflow) and 2) the type of data (e.g. level 2 vs level 3?) and creates a dataset with the expected files.

Additional context
The Python client currently doesn't expose this yet, since datasets are fairly new. However, this should be possible with the lower-level REST API.