[Modelzoo] WDL rebuilding to update API & enable DeepRec features.

[Duyi-Wang]

Rebuild WDL to update API & enable all of the DeepRec features. TODO:

• ~stand-alone model~
• ~replace session with MonitorTrainingSession using another dataset method.~

• enable following DeepRec feature: 7) ~StarServer~ 8) Auto Micro Batch 9) ~FusedEmbedding API~ 10) ~Smart Stage~ 11) ~Auto Graph Fusion~ 12) ~CPU Memory Optimization:START_STATISTIC_STEP, STOP_STATISTIC_STEP, jemalloc~ 14) ~AdamAsync Optimizer~ 15) ~BF16~

  Feature Doc: 1 ) ~Embedding Variable~ 7) ~GRPC++~ 13) ~Incremental Checkpoint~ 14) ~AdagradDecay~

  （optional, default to closed.) 2) ~EmbeddingVariable进阶功能：特征淘汰~ 3) ~EmbeddingVariable进阶功能：特征准入~ 4) Dynamic-dimension Embedding Variable 5) ~Adaptive Embedding~ 6) //Multi-Hash Variable 17) ~WorkQueue~

[Duyi-Wang]

Latest TODO:

• Set Adam as the default optimizer which get a lower perf in stock tf
• ~Set embedding fusion default to open(waiting related PR)~ open following feature pending on Ali:
• Auto Micro Batch https://github.com/alibaba/DeepRec/issues/127
• ~Smart Stage https://github.com/alibaba/DeepRec/issues/122~
• ~Auto Graph Fusion https://github.com/alibaba/DeepRec/issues/126~
• Dynamic-dimension Embedding Variable (Can't find related API in DeepRec!) Not enabled in model: Multi-Hash Variable(DeepRec doesn't provide related API for feature column)

[Duyi-Wang]

Gsteps/s		WDL		Rebuild/old
		value	percent	percent
Old	Community TF	31.78435	baseline	baseline
	DeepRec FP32	33.09086	104.11%	baseline
	DeepRec BF16	47.52758	149.53%	baseline
Rebuild	Community TF	32.5392	102.37%	102.37%
	DeepRec FP32	34.0272	107.06%	102.83%
	DeepRec BF16	48.11835	151.39%	101.24%