Retry on 429s and equivalents

[ozzieba]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
We are trying to do bulk ingestion on a LanceDB table in GCS, and hitting 429 "TooManyRequests" errors, which are not being retried.

Describe the solution you'd like
I'd like 429 responses to always generate a retry with truncated exponential backoff.

Describe alternatives you've considered
This could theoretically be handled by users of this library like LanceDB or even application code. But it seems to me that the best place to address is here.

Additional context
GCS guidance:

You should retry your requests with exponential backoff when:

• Receiving errors with 408 and 429 response codes.
• Receiving errors with 5xx response codes.

In AWS this is a 503 with SlowDown as the reason: https://docs.aws.amazon.com/AmazonS3/latest/userguide/optimizing-performance.html

[ozzieba]

@tustvold I see that you worked on retries in the past (#78), any thoughts?

[tustvold]

Seems reasonable to me, and should be relatively straightforward to implement

[criccomini]

Codex'ed a quick PR #426

[teije01]

Great to see this implemented! I was looking for this in the documentation, but I could not find it listed in the RetryConfig. Perhaps this could be informative to other users too!