In multi-thread condition, EnableProfiler can
be called after RecordEvent is constructed. In this
case, RecordEvent constructor will not init anything,
but RecordEvent destructor will do something since EnableProfiler
was called.
This PR fixes it.

- Added hash function inside of MKLDNN softmax op to be used as handle for primitives stroing in a
context

- Style fixes to softmax mkldnn op

- Fixes after review

- Coding style

- Fix to style

- style fixes

- style fix

- style fixes

- Fix to cody style check

- Rephrasing a comment

* Add FP16 option in save_combine_op

* Fix casting issue

This reverts commit 0446220e.

Reason:

Rerun failed test hides flaky test.
Flaky test can be bugs, for example, race condition.
Test shouldn't be flaky, if a test is flaky, it should be fixed.

* CI: rerun failed tests.

* fix check style error.

* init refactor

* init

* update some comment

* fix build

* fix errorrr

* fix bug

* fix comment

* update

We need to figure out if tensorrt
use row-major or col-major for tensor
layerout inorder to do conversion.

Do not use ctor

* Reduce line of codes.
* We can use virtual function for Maker now.
* The implementation does not care what maker holds, it is easier to
refactor later.