imp fusion_conv_add_prelu and fusion_conv_add_add_prelu op

# Conflicts:
#	src/framework/program/program-optimize/node.cpp
#	src/operators/fusion_conv_add_add_prelu.cpp
#	src/operators/fusion_conv_add_add_prelu_op.h
#	src/operators/fusion_conv_add_prelu_op.cpp
#	src/operators/fusion_conv_add_prelu_op.h
#	src/operators/kernel/arm/conv_add_add_prelu_kernel.cpp
#	src/operators/kernel/arm/conv_add_prelu_kernel.cpp
#	src/operators/kernel/central-arm-func/conv_add_add_prelu_arm_func.h
#	src/operators/kernel/central-arm-func/conv_add_prelu_arm_func.h
#	src/operators/kernel/conv_add_add_prelu_kernel.h
#	src/operators/kernel/conv_add_prelu_kernel.h
#	src/operators/math/gemm.cpp
#	src/operators/math/gemm.h
#	src/operators/math/math_function.cpp
#	src/operators/op_param.h
#	tools/op.cmake

update unit test and jni

update closed #815

Add template for op param

Assign value for scale pointer. close #810

update jni with lock and exception logics && add java interface close #807

add initialization of padding parameter in feed op closed #803

move "temp = numel / (dim[0] * dim[1]);"  out to improve prelu performance 

dd preluop support genet && add unit test  close #799

update register list of resnet closed#797

Add debug info for FPGA track close #795

#793:fix tensor print bug on FPGA