* Convolution fusion operator.
* Clean code
test=develop

The allocator.h header file is required for C-API inference applications

test=develop

* fix dist deps test=develop

* update test=develop

* update test=develop

* update test=develop

* update test=develop

fix compile error on windows

* wip simplify operator framework

* wip

* wip

* done test=develop

* clean test=develop

* fix test=develop

* fix deps test=develop

* fix cpu build test=develop

* fix tensorrt build test=develop

* fix tests test=develop

* fix test=develop

* fix cpu build test=develop

* fix inference on gpu out of mem

the transfer logic in operator.cc will keep creating new scopes.

* Split cichecks

test=develop

* feat(Anakin): expose anakin options to paddle cmake option

Expose ANAKIN_BUILD_FAT_BIN, ANAKIN_BUILD_CROSS_PLANTFORM to Paddle cmake option

test=develop

test=develop

test=develop

test=develop

Fixed problem with array subscript is above array bounds in MKL-DNN jit_uni_reorder_utils.cpp:prb_simplify function

test=develop

test=develop

test=develop

1. Non utf-8 characters within comments of OPs may lead to protobuf fail to parse_from_string
2. comment out some ops which not supported on windows
3. cuda libs may not be correctly linked to target on windows

test=develop