we run into the same issue.
RFC 6585  defines a nice way to inform user agents about a rate
limit: return "429 Too Many Requests", and specify in the Retry-After
Header how long the user agent has to wait before requests will succeed
For now we'd be happy to know the exact rate limit, so that we can
tune our request rate.