ID strategy?

UUID has been brought up a couple times. Both as an enabling piece for an "extraData" approach as well potential data migrations (for pulling together implementations, SaaS, etc)

Using UUIDs is the recommended approach, however we should acknowledge that other unique data elements sometimes need to be enforced. e.g. a Product code should always be unique - ie for a given implementation.

Therefore we should generate an implementation ID which all data is associated to, and then associate a User/api token/seed task with a specific Implementation id.