1.7.0 Release

Highlights

Adds the seed and random state and sampling features to AMPL. (#344) The features are:
- Imbalance-learn sampling
- Seed for Reproducibility
Changes to control model sparsity and improvements to MultitaskScaffoldSplitter (#331):
- Sped up MultitaskScaffoldSplitter and changed its implementation to allow better optimization of validation & test set difference from training set.
- Added split_diagnostic_plots module for visualizing aspects of split quality.
- Added L1 and L2 penalty parameters to XGBoost models to control model sparsity.
- Added hyperopt search domain parameters for NN and XGBoost model sparsity parameters.
Resolved a bug in Transformers fitting where the Transformers for normalizing inputs, outputs, and weights were trained on the entire dataset instead of only the training dataset, potentially causing data leakage. (#385)
Added MODAC API Client + Example Docs (#361)
Incorporates CodeCov into the CI/CD pipeline to generate code coverage reports for enhancing code quality. (#372, #373)

Fixed a bug when running predictions on classification models with balancing weight transformers requires MinimalDataset weights for the prediction data. Previously, get_multitask_perf_from_files_new returned NaN metrics for single-task models mixed with multitask models; now, both return correct metrics. (#387)
Allows the users to combine calculated features with embedded features from pre-trained models (#395)
Logs exceptions generated during a HyperOpt search. Previously they were swallowed, ignored (#392)
Add compute_drug_likeness function to the RDkit_easy module to compute various drug-likeness criteria for compounds in a data frame (Lipinski rule of 5, Ghose and Veber filters, QED), along with the descriptors used to derive them. (#384)
AD calculation improvements. Fixed an error in the calculation. Added the ability to query for the nearest training set neighbors of each compound running predictions for. (#378)
Integrates the MODAC unit tests and automates their execution on GitHub CI/CD Actions (#371)
Implemented unit tests for plotting packages using matplotcheck and the PlotTester API to perform plot validation. (#394)

System clean up:
- Improved the CI test pipeline to eliminate duplicate job executions.
- Added markers to indicate the resources used in certain tests.
- Expanded the range of tests executed in the CI pipeline. Only exclude those that require LLNL resources. (#393)

Correct the AD index calculation for Mordred features containing NaN value (#390)