Refine/op/lstm

add unitttest for mac on ci after some untest being disable

* Implement slice grad operator. #8130
* test slice grad operator and bug fix
* Fix pre commit style

Add error info & remove data sharing between input and output in rnn_…

* Create a new paddle build option to build documentation library

* only build copy_paddle_pybind

* Update the to build paddle_python

* build everything

* Use DWITH_FLUID_ONLY

* add build lite

* add a smaller lib

* Update the gen_doc_lib function to handle multiple cases.

* Prevent Paddle triggering documents deployment.

simple fix

Use memcpy to rewrite the sequence pooling LAST and FIRST mode

add error_info for prelu_op

* hotfix for PaddleTensor buffer.

fix text_classification download error

fix redundant args of lambda and remove exception of destructor

add analysis vis ut

fix mac compile error in subgraph_splitter

* Enhence generate_proposal_labels_op
* Fix bug in generate_proposals_op
* Refine rpn_target_assign_op.
* by Bu Xingyuan, Wang Guanzhong and Dang Qingqing

move bcast op into graph passes

fix fluid_benchmark resnet lr decay

Fix tensorrt subgraph bug

add multi-thread for nlp unit-tests

Support grouped convolution layer with mkldnn.