Citrine

**What’s New**
* To make it easier to keep track of your projects, we've added the ability to search projects in a [ProjectCollection](https://citrineinformatics.github.io/citrine-python/reference/citrine.resources.project.html?highlight=project#module-citrine.resources.project) by name and/or description.
* We've added an example of preprocessing and postprocessing with ExpressionPredictor objects to our documentation.
* Information about whether or not experimental features were used is now available in modules and workflows under `experimental` and `experimental_reasons`.

**Improvements**
* You can now search through projects matching a desired name and/or description by exact or substring match to find a particular project. You do not have to search through all of your projects anymore.

Improvements
* Bounds are no longer required on object templates and are no longer automatically copied from the corresponding attribute template. There is now a distinction between bounds which coincidentally match those of the linked attribute template and not specifying bounds at all.
* In order to preserve list length and ordering, values of `None` are no longer removed during serialization.

What's new
* The GEM table builder now has a new `XOR` (exclusive OR) variable option.
* The model report now includes predictor name and uid and model type.

Improvements
* citrine-python GEMD models now have the same uid auto-generation behavior as gemd-python GEMD models.
* The `register_all` method on datasets is now substantially faster.

Fixes
* The `IngredientQuantityInOutput` GEM Table variable now serializes as the correct type (previously it was serialized as `IngredientIdentifierInOutput`).
* Files with capitalized file extensions (e.g. `.JPG`) no longer fail to upload.

What's New
* We've added a new `register_all` method. This should substantially improve the speed of loading large datasets by substantially reducing network latency. While the implementation of this method is likely to evolve as we gather data on its use patterns and necessary limits, it's a very exciting step forward!

Fixes
* We've moved from building based on GEMD 0.7 to GEMD 0.8, and with that updated how our code consumes that library
* We've eliminated references to a library that was causing some trouble with the latest versions of Python.

What's new
* Registering a dataset now joins the other data collections in supporting dry run validation!

Improvements
* Improved exception handling for better reliability when making remote API calls.
* Clarified documentation on how to build an Ara definition and consume the resulting table in a Data Source.

Fixes
* Updated documentation to reflect new term "coverage probability" instead of "standard confidence".
* Edited minor typos in documentation.

Fixes
* Some requests are now wrapped with exception handling so that when a connection resets, the original request is retried.
* Ingredient Fractions were added to the Predictor modules
* Defaults for the Molecular Structure Descriptor are now handled correctly