Copulas

This release makes Copulas compatible with Python 3.9! It also improves library maintenance by
updating dependencies, reorganizing the CI workflows, adding pip check to the workflows and
removing unused files.

General Improvements

* Add support for Python 3.9 - Issue[282](https://github.com/sdv-dev/Copulas/issues/282) by amontanez24
* Remove entry point in setup.py - Issue[280](https://github.com/sdv-dev/Copulas/issues/280) by amontanez24
* Update pandas dependency range - Issue[266](https://github.com/sdv-dev/Copulas/issues/266) by katxiao
* Fix repository language - Issue[272](https://github.com/sdv-dev/Copulas/issues/272) by pvk-developer
* Add pip check to CI workflows - Issue[274](https://github.com/sdv-dev/Copulas/issues/274) by pvk-developer
* Reorganize workflows and add codecov - PR[267](https://github.com/sdv-dev/Copulas/pull/267) by csala
* Constrain jinja2 versions - PR[269](https://github.com/sdv-dev/Copulas/pull/269/files) by fealho

This release improves performance by changing the way scipy stats is used,
calling their methods directly without creating intermediate instances.

It also fixes a bug introduced by the scipy 1.7.0 release where some
distributions fail to fit because scipy validates the learned parameters.

Issues Closed
* Exception: Optimization converged to parameters that are outside the range allowed by the distribution. - Issue [264](https://github.com/sdv-dev/Copulas/issues/264) by csala
* Use scipy stats models directly without creating instances - Issue [261](https://github.com/sdv-dev/Copulas/issues/261) by csala

This release introduces conditional sampling for the GaussianMultivariate modeling.
The new conditioning feature allows passing a dictionary with the values to use to condition
the rest of the columns.

It also fixes a bug that prevented constant distributions to be restored from a dictionary
and updates some dependencies.

New Features

* Conditional sampling from Gaussian copula - Issue [154](https://github.com/sdv-dev/Copulas/issues/154) by csala

Bug Fixes

* ScipyModel subclasses fail to restore constant values when using `from_dict` - Issue [212](https://github.com/sdv-dev/Copulas/issues/212) by csala

This release introduces a few changes to optimize processing speed by re-implementing
the Gaussian KDE pdf to use vectorized root finding methods and also adding the option
to subsample the data during univariate selection.

General Improvements

* Make `gaussian_kde` faster - Issue [200](https://github.com/sdv-dev/Copulas/issues/200) by k15z and fealho
* Use sub-sampling in `select_univariate` - Issue [183](https://github.com/sdv-dev/Copulas/issues/183) by csala

General Improvements

* Use `corr` instead of `cov` in the GaussianMultivariate - Issue [195](https://github.com/sdv-dev/Copulas/issues/195) by rollervan
* Add arguments to GaussianKDE - Issue [181](https://github.com/sdv-dev/Copulas/issues/181) by rollervan

New Features

* Log Laplace Distribution - Issue [188](https://github.com/sdv-dev/Copulas/issues/188) by rollervan

General Improvements

* Support Python 3.8 - Issue [185](https://github.com/sdv-dev/Copulas/issues/185) by csala
* Support scipy >1.3 - Issue [180](https://github.com/sdv-dev/Copulas/issues/180) by csala

New Features

* Add Uniform Univariate - Issue [179](https://github.com/sdv-dev/Copulas/issues/179) by rollervan