Support fp16 data type for more operations and fix the minor errors.  (Don't use this! It's an incomplete version)

1. Merge with Huchi branch (replace all requireLink with enableGrad that allows gradient computation for a tensor);
2. Update the global memory size (This may make the memory size a little bit larger than the old version).

2. Add MemManager class to manage the memory;
3. Use the new version XList;
4. Add a new interface to create Tensor;
5. Apply MemManager to FNNLM.

1. add macro to implement unary function 2. add sub and div function 3. merge code with the latest branch of xiaotong-working