Pygmtools

What's Changed
* Fix the bug that customized dataset parameters cannot be passed to datasets
* Revise parameters passed to datasets. Now the class and data directory can also be customized

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.9...0.2.10

What's Changed
* Support Neural Graph Matching solvers (PCA-GM, IPCA-GM, CIE, NGM) in PyTorch backend
* Rewrite and reorganize documentation
* Fixed doc compiling issue

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.8...0.2.9

What's Changed
* Fix cropping operation on SPair71k dataset

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.7...0.2.8

What's Changed
* Add progress bar when downloading datsets
* Sort WillowObject data list to facilitate partitioning the training and test set

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.6...0.2.7

* Update documentation for classic solvers and multi graph solvers
* Support non-batched inputs: ``rrwm``, ``ipfp``, ``sm``, ``utils.build_aff_mat``, ``utils.compute_affinity_score``

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.5...0.2.6

What's Changed
* Jittor support for classic_solvers.py by wzever in https://github.com/Thinklab-SJTU/pygmtools/pull/3
* Support black-box differentiation for GAMGM
* End of support for Python 3.6

**Full Changelog**: https://github.com/Thinklab-SJTU/pygmtools/compare/0.2.4...0.2.5