[inferno-ml-bedrock] Add promptWith and related types/functions for structured LLM response

[ngua]

• Adds new JSON and Schema modules for consuming and specifying shape of incoming data from LLMs
• Adds promptWith function for use with said structured LLM responses
• Adds some missing things from Option module to make this usable (flatMap and traverse; for the former, we already had join but flatMap [bind] is nicer to use IMO)
• Adds some tests
  • Note for JSON tests, we will need to wait until another PR since we cannot construct them from Inferno code, they're opaque and are from incoming data only