optimize gemm

增加可被x86编译运行的量化脚本,并增加量化脚本说明文档.

add new param definitions for FPGA ops

update FPGA ops declaration in types h and cpp file

Unify FUSION_CONVADDRELU_OP usage

fix android toolchain

fix #633 modify test cmake

fix #631

fix #628 modify ios build cmake tool chain

modify cmake files for FPGA track

 add FPGA ops declaration and key list

fix #620

Update to the latest one.

add quantification tool to compress binary size

delete unused cmake variable DARMV7