optimize executor.run

On k40 with 4 devices, time reduces from ~4.0 to ~3.8+, should be
more obvious on better hardware

Single device on se-resnet reduce from 0.56 to 0.50

* Fix old documentation for channel_recv

* Initial design of CSP select

* Redesign channel implementation for Select Op

* Remove unecessary header

* Initial checkin of select op, currently will read all the conditional_op in the cases block and also pull out all channels involved in the select.

* Init python select op API

* Python select bug fix when checking op creates block

* Add case_to_execute as (a) input to select, (b) into the passed inputs into the select op

* Add in addition code for select op

* Init fibonacci test from python

* implement fibonnaci sequence test

* update fib unit test

* Improve select test cases

* Shorten non-pep-8-ed lines

* Add methods on channel needed by select op

* Fix compile issues, finish implementation, still need to debug code

* Fix issue with fibonncci test, it works now!

* Change QueueMessage callback to take in an ChannelAction enum, fix select unit test

* Fix case attributes

* Fix issue with select control flow

* Make cases - previously on each selectcase conditional_block - attributes to select

* Use class constants for type of channel

* Change select op to take in "cases" attribute

* return boolean from select callback function to tell Channel if this RECV or SEND should be executed

* Improve attributes and inputs comments on select op

* Fix issues with python unit test

* Assert fibonacci final output

* Fix issue when channel name / channel var is null for "default" case in select op

* Assert base select test output

* Make QueueMessage use shared pointer and modify the order of the callback

* Fixing the order in which the callback is called

* Move channel utility methods to paddle/fluid/operators/concurrency/channel_util

* Create channel_util and move channel util methods

* Fix crash when calling select_op

* Fix deadlock

* Fix issue of channel destructor deadlock

* Fix precommit issues

* Accidentally checked in changes to beam_search_op, reverting change.

* Fix dependency issue in concurrency cmake

* add device_context dependency for concurrency target

* add memopt level

* add opt level for image classification demo

* clean code

* add delete op

* clean code

* test machine translation demo

* clean code

* clean code

* skip fill constant with force cpu

* clean code

* clean code

* refine code

* clean code

* fix bug

Make vec<Tensor> can be serialized to RecordIO

* test cpu float16 data transform

* add isnan etc

* small fix

* fix containsNAN test error

* add data_type transform GPU test

* add float16 GPU example

* fix error

* fix GPU test error

* add context wait

* Redesign channel implementation for Select Op

* Remove unecessary header

* Remove unnecessary comments

* fix is_test issue

* add paddle enforce

* fix bug

* add new func

* small fix

* address comments

* add the type of tuple

* add lod_tensor to tuple

* Adding Python boilerplate code for Go op

* Add very basic test case

* Adding the python logic for go routine

* Fix syntax

* Changing test to notest

* Rename Routine to Go

* Combining GoGuard and Go in one class

* Modify test

* Adding fluid close channel

* Fixing __init__.py for calling fluid.go()

* Adding stubs for channel methods and updating test case

* Removing import *

* Adding imports from concurrency

* Initial commit of GO_OP (for varun)

* Creating local scopes and go through them

* Updated go op inputs persistability enforcement

* Add thread execution; compile failing though

* Fix go op

* Cleaned up Go op

* Fix yapf format issue

* Readd warp ctc dir for unit tests

* Updated make_channel, channel_send, channel_recv and channel_close

* Moved thread function to another method, update unit tests

* remove output var

* Add stubs for channel operators

* Updating concurrency with signatures

* Updated the signature with return status

* Fixed dtype in variables

* Updating stub of ChannelSend + add infershape

* Updating stub of ChannelRecv + add infershape

* Updated signature

* Adding the channel_create operator

* Merge channel send+receive ops

* Update concurrency tests using all operators

* Updating the create op with ChannelHolder

* Fix issues with channel_create_op

* Add the implementation for channel_close op

* Add channel close operator, fix channel close op

* Adding the channel_send op

* Comment channels C++ and Python code

* Concurrency python api comment fix

* Update unit test to add Status variable

* Adding channel receive operator

* Update concurrency test to demonstrate a complete CSP flow

* Fix clang-format issues

* Fixed "Out" parameter name

* Fixing merge conflict in framework.py

* Add channel ops to framework.py no_kernel_op_set

* Seperating channel_send and channel_recv operators

* Documenting capacity type

* Update concurrency test to create go block as child block of main program

* Changing set status implementation

Dim<0> is for scalar (rank-0 tensor). Adding Dim<0> can simplify
a lot of code.