Add option for gRPC client connection retries

[hughevans]

Add a config option for a basic gRPC connection retry policy:

config.connection_options = {
  retry_connection: true
}

That uses a basic policy:

{
  retryableStatusCodes: ["UNAVAILABLE"],
  maxAttempts: 3,
  initialBackoff: "0.1s",
  backoffMultiplier: 2.0,
  maxBackoff: "0.3s"
}

There’s also a retry_policy option to set a custom policy if you want more control:

config.connection_options = {
  retry_policy: {
    retryableStatusCodes: ["UNAVAILABLE", "INTERNAL"],
    maxAttempts: 5,
    initialBackoff: "0.1s",
    backoffMultiplier: 2.5,
    maxBackoff: "0.7s"
  }
}

[DeRauk]

@hughevans thanks for the PR! It looks good overall.

For maxAttempts, initialBackoff, backoffMultiplier, and maxBackoff - I think these are sane defaults, but if we're going to support retries I'd prefer that users be able to pass in their own values to override these if they want. Could you make them configurable?

[hughevans]

@DeRauk Thank you for checking this out. I’ve gone ahead and added a retry_policy option there too.

Also I looked at the README, but was a bit unsure where exactly documentation for this should go logically. Is that something you could take on after merging possibly? 🙇

[DeRauk]

Thanks, and sure.

[DeRauk]

@hughevans typo with retry_policy:)?

[hughevans]

Ah, sorry that only works in newer versions of Ruby. I have pushed a fix now thanks.