Pandera

This release adds two new features to `pandera`.

Improved error reporting

Now failure cases in column checks are displayed in a much more compact format,
where the failure cases, the index of the dataframe where those failures occur, and the
count of failure cases are shown to the user, e.g.

failure cases:
index count
failure_case
foo1 [0] 1
foo2 [1] 1
foo3 [2] 1


Coerce option in `DataFrameSchema` and `Column`

Now the user can `coerce` the dataframe when calling `schema.validate` so that
the columns are cast into the expected data-type before performing `Check`s.

Release Notes

- **Major change**: This release updates changes the API of the `DataFrameSchema` object.
Instead of passing a list of `Column`s, you now pass a dictionary where the keys are `column_name`s
values are `Column` objects. This makes the API feel a lot more familiar for pandas users, who may
often define `DataFrame`s in a similar way (see [README](https://github.com/cosmicBboy/pandera/blob/master/README.md) for details).
- renamed `Validator` to `Check` for brevity and clarity (accordingly renamed `validator_{input, output}`
to `check_{input, output}`.
- created convenience variables for `PandasDtype` so they can be accessed in `pandera` namespace:
`Bool`, `DatetTime`, `Category`, `Float`, `Int`, `Object`, `String`, `Timedelta`

initial release of pandera. API likely to change