Add multiple forests to a database via one file

[rjrudin]

As an ml-gradle/ml-app-deployer user
I want to TITLE
So that I can automate the addition of forests to an existing cluster with as little effort as possible

• Currently, deploying a database is limited to creating every forest based on a single file, and the generation of replica forests is automated
• Need to give a user precise control over adding forests so that e.g. 3 nodes can be added, and replicas for each of those forests will only be added to the 3 nodes being added so that existing nodes are not disturbed

Scenario

Given an existing ML cluster and an ml-gradle setup
And given one or more JSON files in src/main/ml-config/forests/(name of database)/ that contains a forest payload or an array of forest payloads
When the user runs "mlDeployForests" (or "mlDeploy")
Then forests are added based on each directory under forests

• For now, we'll just support adding, unless it proves that updating a forest is easy too (IIRC, I ran into some issues with trying to update a forest via the Management API)
• A single file under ./forests/ will still be supported for the simple case where all the forests can be created the same way
• But now ml-app-deployer will process each child directory of ./forests/, and for each one that has a name matching that of an existing database, the forest files in the directory will be processed. Forests that already exist won't be touched, and forests that don't exist will be added.
• And as mentioned above, a file can be a single forest payload or an array of payloads