Emily

++++++++++++++++++

**Bugfixes**

- Fixed infinite loop in sessions.py when source is DYNAMODB and session ID didn't exist.
- Adjusted tests for pytest to reflect recent session management changes.

++++++++++++++++++

**Bugfixes**

- Main portion of Emily class tried to get session variables before user had session variables, which returns '{}'. When this happens, the default_session_vars and starting_node settings are ignored. Added default_session_vars parameter to sessions.get_session_vars() so that when a session doesn't exist yet, Emily creates a new one with the default session variables.
- self.already_started in the Emily() class was not initialized under some circumstances. Now self.already_started begins as True, and is set to False if Emily is not already started.

++++++++++++++++++

**Bugfixes**

- Allowed for preferred session IDs in 1.0.5, which means session IDs may be strings or integers, but Emily was still trying to cast session IDs as integers.

++++++++++++++++++

**Improvements**

- Support for preferred session IDs when posting to /get_session or calling the get_session() function in the Emily() class.
- Introduced templates for hosting Emily in AWS Lambda and Beanstalk, and for having Emily interact with Amazon's Alexa.

**Bugfixes**

- get_session_vars had a bug when using DynamoDB as the source for session variables.

++++++++++++++++++

**Bugfixes**

- Emily's default log file is 'emily/log/emily.log', but the empty log directory was not included in the wheel. Added log directory path to MANIFEST.in
- Escaped single quote in README.rst that caused the syntax highlights in the command line example to look off.

++++++++++++++++++

**Improvements**

- Added function to setup.py for removing links in README.rst when creating the long description. PyPI does not display the description properly when there are links.

**Bugfixes**

- Forgot to make "user_input" change to Emily's stateless function in release 1.0.2