Treehugger

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

* Update Dependencies
* Drop support for Python 2
* Remove coding header and __future__ imports
* Remove use of six
* In setup.py, remove use of codecs.open, stop requiring kwargs-only and six
* In README.rst, update to "Python 3.4+ supported."
* In requirements.in, remove futures mock and six
* In tox.ini, stop testing on Python 2

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

* Add a ``TREEHUGGER_DATA`` environment variable which points to a YAML file on S3
and reads in the contents. This can be used on systems with out EC2 UserData
support such as containers.

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

* Add support for the ``include`` keyword to point to a YAML file on S3 to
merge in. This can be used to get past the EC2 User Data limit.

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

* Add `--ignore-missing` argument to `exec` command to allow it to run when
user-data isn't configured for the current EC2 instance

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

* Fix issue where ``!!binary`` tags were generated for encrypted variables on
Python 3, whilst they weren't on Python 2. Treehugger should use only strings
in its files for everything now. This may be backwards incompatible for some
users - you should only need to re-encrypt your files for this to work.

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

* Fix a bug where cached values in 'edit' would be reused when changing the
app, stage, or variable names, which change the KMS encryption context.