Adopt persistent storage for various stuff: ResultStorage, RefreshTokenStorage

[phvu]

and probably for Lineage in the future.

The number 1 candidate is MongoDB, but need to be evaluated, e.g. against MySQL. Or redis for caching-related stuff? (Dataframe metadata, etc...)

[phvu]

Copied from the closed issue:

Things that need to be persisted:

• Refresh Token (http-server)
  • String -> Object
  • No cache needed
  • Table with schema (ID, selector, session, etc...)
• ResultStorage: (http-server)
  • UUID -> JSON object
  • cache (small size)
• Dataframe metadata: (dataframe, dataframe-spark)
  • UUID -> object
  • cache (big size)
• Dataframe Lineage: (dataframe, dataframe-spark)
  • key -> object

Tools: MongoDB, MySQL Guava LoadingCache backed by those two