Add support for Sift Science's version-204 API.

[jburnim]

This PR:

• Adds support for calling multiple versions (203 vs 204) of the Sift Science API, where a user can specify which API to use at the client level or in individual method calls.
• Adds support for calling the v204 API, with new parameters abuse_type, abuse_types, and return_workflow_status.
• Adds support for the Workflow Status API, User Decisions API, and Order Decisions API.
• Bumps the version of sift-python to 3.0.0.0, as the client now calls version 204 of the Sift API by default.
• Tweaks the testing infrastructure.

@nickhurlburt -- could you please review these changes?

cc @atpaino , @fredsadaghiani , @yschatzberg

[jburnim]

Looks like test discovery was not added to the unittest module until Python 2.7. I'll investigate switching to unittest2 or only running the tests under 2.6 and 3.4.

[nickhurlburt]

LGTM. What are your current thoughts on supporting Python 2.6 with respect to unit tests?

[jburnim]

What are your current thoughts on supporting Python 2.6 with respect to unit tests?

I updated the Travis testing config to use unittest2. Tests now pass for Python 2.6, 2.7, and 3.4: https://travis-ci.org/SiftScience/sift-python/builds/144594468 .

[nickhurlburt]

Awesome! Looks ready to go, then.

[jburnim]

cc @aaronTufts

[jburnim]

@nickhurlburt , @aaronTufts -- I updated this PR to add support for the return_worfklow_status Events API parameter, as well as the Workflow Status API, User Decisions API, and Order Decisions API.

[nickhurlburt]

Should workflow and decision methods have unit tests?

[jburnim]

Should workflow and decision methods have unit tests?

Tests have been added.

[nickhurlburt]

Still LGTM, thanks!