use yaml files for dataset definition files

[dkrako]

Description of the problem

Currently, new datasets need to be implemented as abstract python classes. This creates a lot of overhead for dataset curators who only want to add their dataset to our library, but don't have any experience with python. The way our fields are defined in DatasetDefinition is overly complicated for the inexperienced user.

Description of a solution

It would be much easier to just use YAML files for defining the datasets instead of python classes inheriting from DatasetDefinition. YAML files are also very intuitive and human-readable

Minimum acceptance criteria

TODO

Sample Code

TODO

[SiQube]

related to #905