Preemptively compute a dataset (i.e. features from natural language
utterances) and store them in Elasticsearch.  This is useful for use with
training, testing, validating and development machine learning models.

The unit of data is an instance.  An instance set (or just *instances*) makes
up the dataset.

The idea is to abstract out Elasticsearch, but that might be a future
enhancement.  At the moment functions don't carry Elassticsearch artifacts but
they are exposed.

There are three basic ways to use this data:

* Get all instances (i.e. an utterance or a feature set).  In this case all
  data returned from [[ids]] is considered training data.  This is the default
  nascent state.
* Split the data into a train and test set (see [[divide-by-set]]).
* Use the data as a cross fold validation and iterate
  folds (see [[divide-by-fold]]).

The information used to represent either fold or the test/train split is
referred to as the *dataset split* state and is stored in Elasticsearch under a
differnent mapping-type in the same index as the instances.

See [[ids]] for more information.

A *client simple* wrapper for an Elasticsearch wrapper.  You
probably want use the more client friendly [[zensols.dataset.db]].

Exactly like [[zensols.dataset.db]] but use the file system.

Instead of using ElasticSearch, use a rows of a JSON file created
with [[zensols.dataset.db/freeze-dataset]].  The file can be created
by any program since it's just a text file with the following keys:

* **:instance**: the (i.e. parsed) data instance (see [[zensols.dataset.db]])
* **:class-label**: label of the class for the data instance
* **:id**: the string unique ID of the instance
* **:set-type**: either `train` or `test` depending on the set type.