Add element-wise multiplication operator

"suggest braces around initialization of subobject [-Werror,-Wmissing-braces]"

refine MKLDNNLayer logical: move forward and backward to MKLDNNLayer and remove copyOutputInfoToOtherDevice

add concat op with CPU kernel

It is buggy in some Linux because the unique_ptr will be free however
the std::string trying to use that char*.

Moreover, it is no need to demangle for error log by Paddle.
Just use `c++filt` or other shell utilities to do this.