AddWorker logic to parallel ops

port over tensor_op_fusion pass

add fusion support for prebuildOpeartion (TensorOpeartion)

fix review cmts

fix ci

fix Ci

fixci

address ci

ci

- add timeout
- add more test cases

fix CI

address review cmts

implment pause in MapOp, added more to callback

add ds_callback

- Initial drop of Python DSCallback

- Pybind DSCallback

- Pybind DSCallback

added callback to mapOp

- de_pipeline DSCallback

- de_pipeline DSCallback

add test case, segfault for now

fix seg fault

- de_pipeline DSCallback

remove 1 line

update callback test case, now works

use builder class for mapOp callback

- de_pipeline DSCallback

- de_pipeline DSCallback

- de_pipeline DSCallback

better test case

minor fix

add comments and minor clean ups

get rid of nullptr in MapOp, use other flag instead

fix a bug ParseMapOp only takes 1 callback

- Added WaitedDSCalabck

refactor callback param

fix text case incorrect number

- added testing

fix cpp test case

- added testing

- revert back lenet changes

- cleanup test_callbacks.py

- cleanup test_callbacks.py

fix CI stage I

fix CI stage II

fix CI and update epoch counter

- add validation
- add more testing  test_callbacks.py

use random data op to do tests

adjust when to call EpochBegin/End

- add repeat with callback

- addressing reviewers' comments

- docstring and CI fixes

- docstring and CI fixes

- docstring and CI fixes

- rebase with upstream/master

fix cpp test case

fix review comments

addr review cmts, add test case