uplink api semantics

[Geal]

Fix #1494
Fix #1539

The Uplink response must drive the behaviour of the schema downloader. Some errors indicate we should not retry.

[Geal]

@abernix how should we handle differences between the poll interval option passed as command line argument, VS the interval returned by Uplink? Should we use the Uplink one, but then override with the CLI arg if it is passed (and so, not set a default value for that one)

[abernix]

Hmm, I knew we had environment variables, but tbh I'd missed we had the ability to set this on the command line. That is unexpected to me and I'm not sure what the user story was for that since I'd say even environment variable usage would be relatively low.

Anyhow, to answer your question, I'd say the value provided by Uplink should always take precedence over anything set by the user since it's a backoff mechanism that our infrastructure teams can turn on if we need to slow things down. (Also we should make sure we have the minimum 10 second guardrail in place for user-configured values.)

(Frankly, I'd deprecate the CLI flag, but that doesn't have to happen here.)

[Geal]

ok, I'm opening #2016 as a follow up