MOSES+RNN Update 8/12

[Joel Lehman]

I've updated my wiki:
http://www.opencog.org/wiki/Extending_MOSES_to_evolve_Recurrent_Neural_Networks#Current_Progress

Basically, Ben's suggestion of increasing population size seemed
to help, and the LRU cache is mostly in, although some bugs
still remain.

My task here as GSoC winds to a close is to port the hardest
variant of pole balancing, test MOSES+RNN on it, and to
objectively measure if the reductions that I have implemented
are effective in practice.

Joel