Flicker

1. Added new methods `any`, `all`, `get_dtype`.
2. Added pass-through methods `groupby` and `write`.
3. Fixed an issue with `min` and `max` methods. They can now be used for non-float columns.
4. Reduced code repetition which should make it easier to read the source code.
5. Added more doc strings.
6. Added more tests to increase coverage from ~52% to ~61%.
7. Added instructions for contributing.

Coverage improved from ~30% to ~50%.

Fixed bugs in `_keys_udf`, `.join`, and improved `.value_counts`.

In this release, we've added lots of exciting functionality.

1. Methods `min`/`max` and `rows_with_min`/`rows_with_max` (which is like `argmin`/`argmax`).
2. A `join` method that allows adding prefix/suffix and doesn't suffer from duplicated column names (like the original `pyspark.sql.DataFrame.join` method).
3. More UDFs (see `flicker.udf`)
4. `flicker.recipes` that contains a context manager to automatically remove temporary columns
5. Tab-completion for column names in jupyter and ipython terminal (`df["a<tab>` would complete to `df["apple"]`)

Also added Apache License 2.0.