TemporaryAllocator is a singleton used for allocating memory for Cudnn. Since it is a singleton, we can delete it for better performance in memory.

We replace TemporaryAllocator by CUDADeviceContextAllocator and CUDADeviceContextAllocation, which uses stream callback to delete the memory allocated for the stream to avoid singleton.

Also added data_feed_proto to operator to fix CI in CPU compilation

* remove attribute in Allocator::Allocate, test=develop

* fix travis ci error, test=develop

* Revert "Revert "Fix allocator bug""

This reverts commit 174d0d0b.

* Revert "fix travis ci"

This reverts commit 5656fa9f.

test=develop

* add inlined_vector.h, test=develop

* add inlined_vector_test,test=develop

add include headers to fix travis-ci
test=develop

test=develop

test=develop

test=develop

Clean allocation->Deleter

test=develop

remove Free() method in UnmanagedAllocator

Use OO style to rewrite memory allocation.

* 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

* Use Builder Chain to decorate new builders. It is easy to extend
  builders.
* Make graphviz path as a build strategy, not a FLAGS.

* cpplint test and add tesnor_py_test.cc

* Update

* Update