1. add fake_quant fake_dequant op
2. add quant_dequant fuse pass
3. fix bug for passes for arm
4. softmax axis problem

fix conv_bias_relu fuse bug，x86 conv kernel bug

test=develop

* bn pattern

* 1. add conv bn fusion pass(conv2d, depthwise_conv2d)
2. modify the graph to ssagraph
test=develop

test=develop

* add concat op, kernel and unitest test=develop

* enable reshape&reshape2 op and add unit test

* enable scale op and add unit test

* fix XShape checking for reshape2 unit test
test=develop

* remove op_desc.hasInput(..) and op_desc.hasOutput(..) to adapt X86 platform
test=develop

* remove target() and precision() from softmax, reshape, reshape2 and scale op
fix CopyDataFrom() of TensorHvy
test=develop

* alloc memory then copy tensor's data during invoking TensorHvy::CopyDataFrom(...)
test=develop

* fix conflicts

* fix kernel registry realted bugs test=develop

* enable softmax op and add unit test

* move softmax sub-functions to softmax.cc, and move basic math functions to funcs.h

* migrate several ops:
  mean,
  mean_grad
  fill_constant
  square_grad
  elementwise_sub_grad
  mul_grad

* add sdg_op

* fix kernel platform registration issue

* code cleanup

* fix platform typo

test=develop

load_model
add scale, mul Op

and #pragma once in edited headers

add #include<string> in edited headers

test=develop

* link IR graph to analysis graph

* add clean code and update

* add infer_clean_pass

* add ir_pass_manager

* support fc fuse executation

* fix ir circle

Reduce one level of inheritence.

add doc