Dataset API are inconsistent between configuration time and runtime

This is for embedded datasets:

• in DatasetDefinition.configure() the instance name is the unqualified name

• in DatasetDefinition.getAdmin() and getDataset() it is qualified (prefixed) with the name(s) of (the) enclosing dataset(s)

The consequence is that configure() cannot perform any configuration that is based on the instance name. For example, the current FileSet configures its base location based on its name. That leads to incorrect behavior - the embedded file sets of all partitioned file sets share the base path "/files", which is obviously wrong.