* support promote dtype for static amp training

* unify o1 and o2

* update for unittest

* fix op_role

* add use_promote arg

* fix doc

* add promote unittest

* polish unittests

* fix controflow and test

* Unify the static amp codes of fp16 and bf16.

* Polish apis and add unittest.

* Add operator stats collecting tools for program.

* Add the check of number of bloat16 operators in unittest.

* Add warning for operator not supported for amp.

* Add testing of BF16 O1 and O2.

* rem is_compiled_with_npu

* rem nup related code

* make lint happy

* rem test

* remove some tests

* Update grad_scaler.py

* fix an error

* rem is_compiled_with_mlu

* fix some mlu_place and mlu_device_coount

* make lint happy

* [CodeStyle][py2][U004] unecessary explicit `object` inheritance in class definition

* fix an increment

* update config

* re-blacken python code

* temporarily disable date and diff_py_file

* skip a format

* Fix scatter_op fp16 perf problem.

* Add scatter into black list.

* Add scatter into black list for dygraph.

* support FP16 for more ops

* add amp list tests

* refine reduce_mean_grad

* fix OP benchmark ci

* fix fp16 reduce_mean

* updat ut, but still have some problems

* remove mean/reduce_mean fp16 kernel

本PR是fused_transformer的layer层代码，包含FusedFeedForward的layer层代码和FusedTransformerEncoderLayer的代码。

* reduce sum op default fp32, add into amp black list

* reduce_sum default fp32 can avoid return inf when the sum value large than 65504

* Use the runtime to create the unsupported_fp16_list using in AMP.

* Add more infos about supported ops.

* Add some comments for the function of OpSupportedInfos.

* Fix the unit test of test_multi_precision_fp16_train.

* add fleet amp.init()

* add unittest for fleet_amp_init

* add cast ops before and after unsupported fp16 ops.

* Keep partial net in FP32 pattern.

* Support check_finite_and_unscale and update_loss_scaling for FP16 calculation mode.

* Add fp16 support for adam op.

* add multi precision attr for adam.

* Fix the bug of test_multi_precision_fp16_train UT.

* Code format for CI.

* Fix the redefine error about MPTypeTrait on windows.

* fix bugs of the _create_accumulators func in Momentum.

* fix bug when inserting post cast op.

* Add the update_loss_scaling op in allow_set of UnusedVarCheck.

* Update for ci coverage.

* Add some doc for OptimizerWithMixedPrecision.

* Fix the code style.

* Imporve the doc of `amp_init`.

* Change for fp16 testing if users have the infer program defined in separate way.

* add fp16 for layer_norm op

* revert layernorm api

* fix forward

* fix forward

* fix backward for layernorm with fp16

* fix unit test for layernorm with fp16

* fix with_mkldnn compile error for layernorm with fp16

* 1. revert to PADDLE_ENFORCE_NOT_NULL, 2. change static_cast<float> to static_cast<U>

* fix with_mkldnn compile error for layernorm with fp16

* fix with_mkldnn compile error for layernorm with fp16
Co-authored-by: Nzhiqiu <chenqiuliang@baidu.com>

* add matmtl_v2 to amp list

* support dygraph

* add fused_bn_add_relu op

* fix some typos in AMP. test=develop

* delete useless codes. test=develop

* add custom black varname test=develop

* fix dtype test=develop

* fix num test=develop

* fix ut test=develop

* fix coverage test=develop

* fix blackvar names test=develop

Optimize amp for multi-gpu to enable FP16 gradients transfer across gpus

init new amp, optimize inserting cast op for batchnorm