add h2o schema

[ryanSoley]

What

• adds the ability for self artifacts to be logged with Experiment methods other than log_artifact
  • required to use log_h2o_model from log h2o models as artifacts #415
• adding schema for some h2o models
  • H2OGeneralizedLinearEstimator
  • H2OGradientBoostingEstimator
  • H2ORandomForestEstimator
  • H2OTargetEncoderEstimator
  • H2OXGBoostEstimator

How to Test

• python -m pytest tests/integration/test_schema.py

[ryanSoley]

needs #415 merged to log the model artifacts via schema

[ryanSoley]

the h2o model contains a metrics table and feature importances, but they are stored in h2o specific data types. I'll open followup stories to investigate how to get that data logged with the schema too. for now, its all available on the estimator logged with log_h2o_model