Don't honour Retry-After header for non-retryable errors

[tigrannajaryan]

The spec is currently written in a way that suggests Retry-After header should be honour for all HTTP response codes, whether success, failure (retryable or not). I don't think this is right and I don't think was ever the intent. I think we should treat it as a bug.

I suggest that we change this:

The client SHOULD honour the waiting interval specified in the "Retry-After" header if it is present.

to this

For retryable response codes the client SHOULD honour the waiting interval specified in the "Retry-After" header if it is present.

Retry-After does not make sense for non-retryable codes. I think we should consider this a bug. It was never the intent to have a waiting interval for non-retryable response codes. That's pointless. Whoever wrote the phrase (me?) just used poor wording.

I also think the spec does not need to say anything about what happens with Retry-After for non-retryable codes. The client must simply ignore Retry-After in those cases, just like it ignores all other irrelevant headers that may be present in the response.