Meta Dataset integration

[pesser]

From #200 which introduced Meta Datasets:

Things, that should be done in the near future:

* Port the `EvalDatafolder` and the whole eval pipeline to this new dataset class (This should be relatively easy, as large parts of this class are derived from the `EvalDatafolder`)

* Once the eval pipeline is ported, package the data writing routines, such that we can easily reuse them to export arbitrary datasets.

* Write a test for the `MetaDataset`, which uses a custom function as loader

[pesser]

#217 switched eval pipeline to MetaDataset. Next up: resurrect edcache with MetaDataset.