Include smart handling of rate limiting when requested

[ships]

For majority of users, this API will return 429 Too Many Requests once you start to get rate limited. This will last some number of seconds and the limit is generally 50 requests per second.

When creating this Client, with intent to re-use, we should allow you to retry-with-fallback in a standard way that you can configure the delay. For example just wait a second if you get the characteristic error -- 429 Too Many Requests.

[ships]

Iceboxing this because the interface for substituting out request handling logic in the OpenAPI generator library is pretty obtuse. All class-based and the injectable type is the constructor for an abstract class, so there is very little modifiability.