[Modelzoo]Rebuild MMoE to update API and Enable DeepRec Features

[jmaksymc]

WIP

• Enable following DeepRec feature:
  • Enabled By Default: 8) Auto Micro Batch same with https://github.com/alibaba/DeepRec/issues/127 9) FusedEmbedding API, embedding fusion 10) Smart Stage same with https://github.com/alibaba/DeepRec/issues/122 11) Auto Graph Fusion https://github.com/alibaba/DeepRec/issues/144 12) CPU Memory Optimization:START_STATISTIC_STEP, STOP_STATISTIC_STEP, jemalloc 14) AdamAsync Optimizer 15) BF16
  • Disabled By Default: 1) Embedding Variable Not support EV yet 7) GRPC++ and StarServer Implemented in the model but they are not supported yet 13) Incremental Checkpoint 14) AdagradDecay 2) EmbeddingVariable advanced features：Embedding Elimination Not support EV yet 3) EmbeddingVariable advanced feature：Embedding Filter Not support EV yet 4) Dynamic-dimension Embedding Variable - There is no feature_column API 5) Adaptive Embedding - ERROR: data.shape must start with partitions.shape, got data.shape = [18886], partitions.shape = [512] 17) WorkQueue
  • Other Features: 6) Multi-Hash Variable - There is no feature_column API

[Duyi-Wang]

Please update README、distribute training files and add DBMTL in modelzoo/README. Adaptive Embedding issue has been submitted to Ali https://github.com/alibaba/DeepRec/issues/236

[shanzhou2186]

Please help to change the steps number and update the performance data of Alicloud CPX in JIRA and ignore the accuracy for such a small dataset. Our SPR platform ran into issues, which hasn't been fixed yet. Then we can push the PR to Alibaba.

[Duyi-Wang]

• Please add a directory structure at the start of REAMDE.
• Please update distribute training files, keep align with other models.

[jmaksymc]

Updated, performance is on JIRA.

[jmaksymc]

Cleaned comments, changed train/test files names.