Randomly generated actor atoms and molecules now control NAO

https://github.com/ctf20/DarwinianNeurodynamics/commit/fdc9daf921314a7f32fdc374404f9d12bfe82266

The above code in github works out of the box. It randomly generates a population of atoms which call each other to control the NAO. Each atom contains a feedfoward neural network, again randomly initialised with inputs a subset of sensory data from ALMemory, and actions direct primitive motor commands with setAngles to joints.

The next step is to permit other KINDS of atom, that are not primitive action atoms.

28th Feb. SHC atom type included.

https://github.com/ctf20/DarwinianNeurodynamics/tree/master/COND_MODULAR3

Next step: Hand design a visually guided reaching actor molecule that has good functionality AND which can be evolved by mutation of atoms and links between atoms.