Enable sequential execution mode in parallel executor

* remove unnecessary var_and_op
test=develop

* fix _init_zero_idx_
test=develop

test=develop

add depthwise conv mkldnn pass

add program check

* enable pyreader use pin memory

* add py reader pin memory test test=develop

* Add affine grid generator.

* fix ffine grid.

* Add unitest.

* Add CPU kernel and fix unitest.

* Fix CPU kernel.

* Refine code.
test=develop

* Fix python api.
test=develop

* Update python api.
test=develop

* Fix comment.
test=develop

* Rename affine_grid_generator to affine_grid and enhence unitest.
test=develop

* Fix unitest.
test=develop

* fix dim0 in _load_slice_up_vars

* fix dim0 in _load_slice_up_vars, fix innershape in delete_var_op

* Revert "fix lookuptable in reduce strategy"

This reverts commit 0e722c5e

* add unit test for dist

* add unit test for dist, test=develop

* cancel revert, test=develop

Fix the example in the doc of transpose_op

Refine the doc of dynamic_gru and gru_unit.

remove unused WITH_FAST_BUNDLE_TEST option

added transpiler pass for mkldnn depthwise_conv

test=develop

add a small test to verify tensor type

set en empty optimize block if pserver has no optimize block

Dist table support multi table

* fix GetTensorFromVar
test=release/1.1

* refine GetTensorFromVar
test=develop

test=develop

test=develop

update doc to 1.1

add fused broadcast op unit test

test=develop

It seems that the blocking queue might be destroyed early than Run
method complete. It might because the Run method throw some unhandled
exception. However, it should be shared_ptr when multthread access an
resource. So change BlockingQueue as a shared_ptr.

test=develop

fix sequence_pad example error

* fix use_reader_alloc uts

* dist ut fixes test=develop

* update test=develop

* fix test for py3 test=develop

test=develop

remove the optimized but buggy implementation

test=develop

Fix dynamic_gru h_0 bug

Fix Mac Python3 CI job