Kyklo code exercise. Notes:
- Uses sqlite for simplicity.
- Uses a many-to-many relationship between organizations and models, which makes
more sense to me. This requires an API change; specifically, requests have to
be namespaced with the organization name, as in
/:organization/.... - Uses token-based auth. In a real application we would use real user-specific
tokens, but here we just use accept the string
kyklofor simplicity. - Does not do any cacheing of requests to reuters, github etc. In a real application these requests should not be made synchronously during the request-response cycle. They should be handled in a background task, using Active Job or similar.