Citrine

This release introduces a couple new exciting features related to AI models, and continued improvements to documentation and access control.

What's New
* The `MeanPropertyPredictor` now supports featurization of both real and categorical properties in a formulation. The categorical-valued mean property is a quantity-weighted distribution over property values in the mixture. The same interface applies to both real and categorical values, so give it a go in your workflow! 723
* A new capability to convert deprecated `SimpleMLPredictor` nodes to `GraphPredictor`s with equivalent components. The methods `convert_to_graph` and `convert_and_update`, available on a `PredictorCollection`, enable this feature. 720

Improvements
* Expanded search behavior in support of our new Teams Access Control model. 718
* Expanded documentation on default predictors and the new `MeanPropertyPredictor`. 722 723

**Full Changelog**: https://github.com/CitrineInformatics/citrine-python/compare/v1.21.2...v1.23.0

This release is mostly about supporting our new Access Control model, with some improvements in documentation, usability, and testing along the way.

What's New
* In support of our upcoming improvements to our Access Control model, we've added support for Teams. Note that Teams functionality will not be available to every deployment 699 701
* Branches are now available as an organizational concept for design workflows. See the documentation for more information on use cases: [Branches](https://citrineinformatics.github.io/citrine-python/workflows/design_workflows.html#branches) 702 707 708

Improvements
* Positional arguments for the [Citrine](https://citrineinformatics.github.io/citrine-python/reference/citrine.citrine.html?highlight=citrine#citrine.citrine.Citrine) constructor have been deprecated. Hopefully this will help avoid some stubbed toes, as people had an unfortunate habit of passing the `host` as the second argument. 706
* We've added python 3.9 and 3.10 to our travis tests and bumped required versions for pandas and sphinx packages. 713

Fixes
* Multiple corrections were made to the documentation. 704 709 711

**Full Changelog**: https://github.com/CitrineInformatics/citrine-python/compare/v1.16.0...v1.21.2

This release should see substantial speed improvements with handling GEMD objects.

Improvements
* The biggest change is moving to [gemd-python v1.4.1](https://github.com/CitrineInformatics/gemd-python/tree/v1.4.1). This brings along with a number of performance improvements. #696 700

Fixes
* We added a check to help in migrating our account management services. 695
* We cleaned up our test suite so that it doesn't throw any warnings, updating a couple spots that were still using deprecated methods. 697
* We did some minor documentation fixes. 698

This release includes substantial reorganization around data object registration (no breaking changes) as well as minor quality of life improvements.

Improvements
* Thanks to a little reorganization, the `register_all` method on any of the object collections can now accept objects of any type (688). While the library always dispatched object types to the correct object collections, not being able to mix and match meant that `dry_run` couldn't be used to validate a whole material history.
* To help out with large object registrations, we've added a `status_bar` flag to register_all (694). Relax in the soft glow of slowly increasing numbers as you populate the platform.
* We're now using [gemd-python v1.2](https://github.com/CitrineInformatics/gemd-python/releases/tag/v1.2.0) for our data objects (#691). With this, you can now use `==` to compare complex material histories.
* And on the modeling side, we've added `INFER` to the `auto_configure` function (692). This will check your `DataSource` for a `FormulationsDescriptor` by default, which should help protect some toes from being stubbed.

Fixes
* And with thanks to external contributor shenganzhang and [pytest-flakefinder](https://github.com/dropbox/pytest-flakefinder), our tests are now robust to repeat evaluations (#690). Our community makes us stronger.

This maintenance release includes an updated request header, documentation fixes and improved testing.

Improvement
* Added a custom user-agent header that identifies the citrine-python version https://github.com/CitrineInformatics/citrine-python/pull/685
* Dataset deletion prompt now includes the Dataset `name` and `id` https://github.com/CitrineInformatics/citrine-python/pull/682

Fixes
* Fixes to code examples in the documentation https://github.com/CitrineInformatics/citrine-python/pull/687 and https://github.com/CitrineInformatics/citrine-python/pull/683
* Increased test coverage for the `user_agent` https://github.com/CitrineInformatics/citrine-python/pull/686

This is a maintenance release that adds `config`, `name`, and `description` attributes to `GemTable` to improve usability and allow `GemTable`s to be found using the `citrine.seeding` methods. Several improvements for `citrine-python` developers are also included.

Improvements
* Added `config`, `name`, and `description` attributes to `GemTable` to improve usability and allow `GemTable`s to be found using the `citrine.seeding` methods. (https://github.com/CitrineInformatics/citrine-python/pull/681)