Add a few more RecordEvent.
Cleanup

Sometimes dev_ctx is not available when RecordEvent.

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.

* Add profiling information for inference example, recognize digits.

* Refine the profiling method.

* Correct the use of RecordEvent and simplify recognize_digits.