AutoML libraries that use DatasetProfile instead of DatasetInfo

[justinxzhao]

Changes:

• create_auto_config_with_dataset_profile (mirrors create_auto_config), which uses the dataset_profile_pb2.DatasetProfile (whylogs-based) to infer feature types. A few notable API modifications:

  • dataset and dataset_profile are separate parameters, and it's an error to specify both simultaneously
  • tune_for_memory is excluded
  • include_hyperopt is a new parameter that is False by default.

• Rename _get_hyperopt_config to get_resource_aware_hyperopt_config

• Add proto-generated go files, which enables DatasetProfile objects to be interfaced with in go code.

• Moves the hyperopt section of the base_automl_config.yaml to a separate function get_default_automl_hyperopt, defined in-line.

[github-actions]

Unit Test Results

         6 files  ±  0           6 suites  ±0   3h 35m 25s ⏱️ - 32m 27s
  3 539 tests +  2    3 464 ✔️ +  2    75 💤 ±0  0 ❌ ±0 
10 617 runs  +46  10 374 ✔️ +45  243 💤 +1  0 ❌ ±0 

Results for commit 22b627c. ± Comparison against base commit 6153df9.

♻️ This comment has been updated with latest results.

[arnavgarg1]

LGTM! Just left some comments for minor nits