Token issuance should be a pipeline

Token issuance should become a pipeline instead of the current mechanism. This would also involve refactoring the token_api to include a chunk of the logic from the token_provider_api (rather than having disparate subsystems). When this occurs token data should become a single data store and each token "type" (version) created should derive the relevant data from the original data in the token store rather than storing an ever creeping volume of data in the tokens.