Citrine

This is a minor release with a few bug fixes and support for a couple of new Predictor types.

What's New:
- Two new Predictors: ExpressionPredictor and GraphPredictor.
Improvements:
- Default log level is now WARN - your jupyter notebooks should be showing a lot less pink
Ingredient Runs now inherit names and labels from their Specs, and are automatically populated.

Fixes:
- Fixed a case where material histories could be incorrectly deserialized

What's New
- Ability to run a PerformanceWorkflow on a predictor
- Ability to set the `active` flag on a module

Improvements
- Updated documentation
- Added a Dockerfile for running unit-tests
- Improved error messaging for non-2xx http statuses

This forces an upgrade to [taurus v0.3.0](https://github.com/CitrineInformatics/taurus/releases/tag/v0.3.0), which supports a much larger set of units.

What's New
* The [units list](https://github.com/CitrineInformatics/taurus/blob/master/taurus/units/citrine_en.txt) has been greatly extended.

This release adds the ability to set more granular permissions for project members (READ/WRITE) and increases the documentation coverage.

What's New
* Users can set more granular permissions for user's under a project.

Improvements
* Updated documentation.

This release supports Citrine's AI functionality added into the citrine-python client. Users can now create modules and workflow in support of sequential learning to generate material candidates.

What's new
* Registering and validating workflows, design spaces, predictors, and processors. See https://citrineinformatics.github.io/citrine-python/ for complete documentation on all the new concepts that are introduced.
* Training predictor modules and viewing predictor reports.
* Generating and scoring generated candidates against objective and constraints.

Improvements
* Better visibility regarding specific failures and exceptions from the API