Experiment on vgg flower, 2 trainers, 1ps.
    more trainer could have more speedup.

    After:
    Pass = 0, Iters = 327, Speed = (7.52) img/s
    Before:
    Pass = 0, Iters = 385, Speed = (6.77) img/s

* listen_and_serv use localscope

* fix ut

* Fix a bug related to dispensable inputs and refine the inference unittest.

* Fix the use of dispensable inputs in reshape_op.

* Polish the enforce statements.

* Fix an English writing typo.

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.

Add the call of DropKids at the end of executor.Run to delete the local scopes created in operators (#10403)

* Add the call of DeleteScope to delete the memory of scope created by NewScope.

* Call DropKids at the end of executor.Run to delete all local scopes created in operators.