Feature/python

[Chrisgant]

The python interface is now at a point where we can completely generate MarkovChains from python. The purpose of this new feature is simplification of the workflow and accessibility to non power users.

The interface includes exposed classes and functions for:

• Lattices
• Actions
• Integrators
• MarkovState
• HMC

And type casters:

• dict <-> NSL::Parameters
• torch.device <-> NSL::Device
• torch.tensor <-> Torch::Tensor
• dict <-> NSL::Configuration
• string <-> YAML
• string <-> H5Node

An example can be found here.

Due to the amount of templates in our C++ code, some type specifications/simplifications were necessary.