Gerrychain

Quite a few things have changed since the last release, so it's time for a new one.

Some highlights:
* More informative error warnings for invalid geometries (thanks pjrule !) 262
* A `.from_districtr_file()` method for creating a `Partition` using a plan drawn in [Districtr](https://mggg.org/Districtr) #272
* Test coverage is now at 90%
* A `.plot()` method for more easily plotting districting plans 271
* Support for setting a random seed for reproducibility, along with documentation with reproducibility tips
* GerryChain is now licensed under the BSD 3-Clause License! 264

* Switch to GPLv3 instead of AGPL 259
* Support repeatable/reproducible chain runs (with some environment setup) 257

Notable changes:
* Fixes serialization errors in `Graph.to_json()`
* Removes PySAL dependency by implementing our own graph creation logic, which also drastically improved graph creation times

This release fixes some issues in the `Graph` class, bringing it in line with the Getting started guide.

🎉

This is the initial release of the GerryChain package, for generating ensembles of districting plans using Markov chain Monte Carlo. This library is authored by the [Metric Geometry and Gerrymandering Group](https://www.mggg.org) as part of our efforts to support basic research into gerrymandering and empower public participation in the redistricting process.

User feedback (bug reports, feature requests, questions) is welcome! [Open an issue](https://github.com/mggg/gerrychain/issues) in our GitHub repository to start the conversation. Contributions are also welcome in the form of [pull requests](https://github.com/mggg/gerrychain/pulls).