For V0.4: Code Merge, clean up and documentation

[walkerning]

• [x] clean up examples organization.
• [x] clean up unused codes, such as trainer/archnetwork_trainer.py, evaluator/bftune.py, controller/compare.py, controller/archnetwork_controller.py...
• [x] README
• [x] Hardware related profiling & parsing flow. (flow description and example)
• ~Add a Dockerfile~ seems unnecessary for a Python package with clear dependencies
• [x] more primer doc:
  • [x] cfg breakup of ENAS/DARTS
  • [x] orchestration pic, including method calls
  • [x] add DARTS instruction message: python file location, method explain
  • [x] the differences of the saved ckpt, what should be used in derive...
• [ ] RNN example configs, search/final

Research docs

• [x] README/configs/scripts for FTT-NAS
• [x] README/configs/scripts for search space profiling
• [x] README/configs/scripts for GATES
• [x] README/configs/scripts for Surgery

[walkerning]

• [x] I think shuffle_data_before_split should be changed to False... unexpected behavior with different seed (training data accidentally used by controller update/search too), or configuration mismatch.
• [x] other_kwargs support of prepare_data_queues
• [x] async trainer continue development, test with detnas
• [x] hardware interface development

v0.4

• [x] hardware objective model assets
• [x] merge more DetNAS codes