Proglearn

Added Features
1. Include `python` 3.9 compatibility checks
2. Add new experiment & tutorial notebooks

Bug Fixes
1. Fix `black` formatting on notebooks & scripts
2. Resolve dependency conflicts

Documentation Fixes
1. Optimize package dependencies in `requirements.txt`
3. Update algorithm descriptions to `Synergistic Forests (SynF)` and `Synergistic Networks (SynN)`

Added Features
1. `LifelongClassificationForest` and `LifelongClassificationNetwork` now use attributes & functions from `ClassificationProgressiveLearner`.
2. Add tests for transformers, forests and networks, increasing code coverage.
3. `CircleCI` now automatically pushes repo releases to `pip`.
4. Let `UncertaintyForest` inherit from `LifelongClassificationForest`.

Bug Fixes
1. Correct `network_construction_proportion` to `default_network_construction_proportion` in `LifelongClassificationNetwork.add_task`.
2. Update `keras` imports for notebooks.

Documentation Fixes
1. Optimize API references.
2. Update issue templates.
3. Add `black` format checks for benchmarks & notebooks.
4. Add attributes to forest & network docstrings

Bug Fixes
1. Changed from `keras` to `tensorflow.keras`.
2. Remove 2D array check in `NeuralClassificationTransformer`.

Documentation Fixes
1. Separated notebooks into `experiments` & `tutorials`.
2. Optimized API references & removed internal functions.
3. Added `CITATION.cff` for software citations.
4. Updated `Zenodo` badge.

Added Features
1. Switched to CircleCI for code testing.

Bug Fixes
1. Specified `classes` parameter of `TreeClassificationVoter` and `KNNClassificationVoter` to be formatted as `np.asarray`.
2. Relocated the initialization of `lf_` attribute for `UncertaintyForest` to the `fit` function. Allowed `uncertaintyforest_posteriorestimates.ipynb` benchmark to run as intended.
3. Fixed the decider data splitting in the `set_decider` function of `ProgressiveLearner`. Let the decider use its allotted data set by `decider_idx = self.task_id_to_decider_idx[task_id]`.
4. Specified new `numpy` version in `requirements.txt` to both resolve version conflicts with `tensorflow` and maintain support for Python 3.6.

Documentation Fixes
1. Included "Tutorial Guidelines" in `contributing.rst` for better online display.
2. Updated and added badges.

Added Features
1. Added `n_estimators` to LifelongClassificationForests `add_{task, transformer}` functions. This allows each task to have a different number of trees.
2. Previously, the classes in forest.py used `finite_sample_correction` with the only possible value of `kappa` being 1. We added the ability to set any `kappa`.

Bug Fixes
1. Previously, the default values of all model parameters to LifelongClassification{Forest, Network} `add_{task, transformer}` functions were None. The value of None indicated to use the default model parameter specified in the instantiation of the LifelongClassification{Forest, Network}. But, this meant that a user would be unable to train a new tree transformer to purity (by setting max_depth = None) unless `default_max_depth` was None. This is an undesirable restriction. So, we changed the default value of all model parameters to LifelongClassification{Forest, Network} `add_{task, transformer}` functions to the string "default" - this indicates to use the default model parameter specified in the instantiation of the LifelongClassification{Forest, Network}.
2. Updated selection of {transformer, voter, decider} data to be without replacement.
3. Previously, we were encountering issues when the voters across bags would output posteriors of different lengths (caused by only training on a subset of the classes). We manually fixed these to ensure that all voters across bags output the same number of classes, by appending 0's to missing classes in the voter posterior estimates.
4. Type checked all inputs X, y to {network, forest}.py classes.

Documentation Fixes
1. Previously, many of the docs did not render correctly in Sphinx because they were formatted incorrectly. We fixed this formatting error and now all documentation renders correctly in Sphinx (and thus the web docs).

Presentation Changes
1. `transformer_voter_decider_split` was changed to `network_construction_proportion` in network.py
2. Minimized documentation and Python Package to only expose well-documented, tested functions that are described in the paper

Documentation Updates
1. We converted the documentation to be hosted at proglearn.neurodata.io
2. We converted the documentation style to follow that of hyppo (https://github.com/neurodata/hyppo)

Repository Updates
1. We refactored the repository folders to follow the structure of hyppo (https://github.com/neurodata/hyppo)