- Any C++ type with Lua interface generated by Swig and
	  composed into a SwigWrap type can be read via PyTorch
	  with auto-completion support in Python.
	- PyTorchAug import of 'nn' module has been generalized to allow
	  import of Lua classes from any module.
	- Added Boolean to types that can be popped from Lua stack into Python

        - Added luaL_loadbuffer(...) interface to call Lua code
          directly from a string, for given chunkname
        - Swig/C++ objects appear as userdata in Lua along with a
          specially formatted metatable. SwigWrap implements a Lua
          torch.class that carries all functions for a given userdata.
        - The Lua code required for this is embedded in PyTorchAug
        - PyTorchAug.popSomething is extended to automatically convert userdata
          into SwigWrap torch.class;

Remaining TODO

	- check for compatible userdata before type wrapping in popSomething
	- add tests that demonstrate the Lua/Swig interface