Datatest

------------------

* Changed bundled pytest plugin to version 0.1.3:

* This update adds testing and support for latest versions of Pytest
and Python (now tested using Pytest 3.3 to 4.1 and Python 2.7 to 3.7).
* Changed handling for 'mandatory' marker to support older and newer
Pytest versions.

------------------

* Improved data handling features and support for Python 3.7.
* Changed Query class:

* Added flatten() method to serialize dictionary results.
* Added to_csv() method to quickly save results as a CSV file.
* Changed reduce() method to accept "initializer_factory" as
an optional argument.
* Changed filter() method to support predicate matching.

* Added True and False as predicates to support "truth value testing" on
arbitrary objects (to match on truthy or falsy).
* Added ProxyGroup class for performing the same operations on groups of
objects at the same time (a common need when testing against reference
data).
* Changed Selector class keyword filtering to support predicate matching.
* Added handling to get_reader() to support datatest's Selector and Result
objects.
* Fixed get_reader() bug that prevented encoding-fallback recovery when
reading from StringIO buffers in Python 2.

------------------

* Added impoved docstrings and other documentation.
* Changed bundled pytest plugin to version 0.1.2:

* Added handling for a 'mandatory' marker to support
incremental testing (stops session early when a mandatory
test fails).
* Added '--ignore-mandatory' option to continue tests
even when a mandatory test fails.

------------------

* Added a bundled version of the pytest plugin to the base installation.
* Added universal composability for all allowances (using UNION and
INTERSECTION via "|" and "&" operators).
* Added ``allowed`` factory class to simplify allowance imports.
* Changed is_valid() to valid().
* Changed ValidationError to display differences in sorted order.
* Added Python 2 and 3 compatible get_reader() to quickly load
csv.reader-like interface for Unicode CSV, MS Excel, pandas.DataFrame,
DBF, etc.
* Added formal order of operations for allowance resolution.
* Added formal predicate object handling.
* Added Sphinx-tabs style docs for clear separation of pytest and
unittest style examples.
* Changed DataSource to Selector, DataQuery to Query, and DataResult to
Result.

------------------

* Added module-level functions: validate() and is_valid().
* Changed DataQuery selections now default to a list type when no
outer-container is specified.
* Added DataQuery.apply() method for group-wise function application.
* Changed ValidationError repr to print a trailing comma with the last item
(for ease of copy-and-paste work flow).
* Changed sequence validation behavior provides more precise differences.
* Added truncation support for ValidationErrors with long lists of differences.
* Changed excess differences in allowed_specific() definitions no longer
trigger test failures.
* Added support for user-defined functions to narrow DataSource selections.
* Added traceback hiding for pytest.
* Fixed bug in DataQuery.map() method--now converts set types into lists.

------------------

* Added Boolean composition for allowed_specific() context manager.
* Added proper __repr__() support to DataSource and DataQuery.
* Changed DataQuery so it fails early if bad "select" syntax is used or if
unknown columns are selected.
* Added __copy__() method to DataQuery.
* Changed parent class of differences so they no longer inherit from Exception
(this confused their intended use).
* Changed documentation structure for ease of reference.