Implement exponential backoff retry logic

[codebestia]

Description

Transient failures (connection resets, 502/503/504 responses) should be automatically retried with exponential backoff and random jitter rather than immediately surfacing to the caller. This makes the SDK resilient to brief network hiccups.

Proposed Steps

• Implement a _retry_with_backoff(fn, max_retries, base_delay) utility.
• Retry on: httpx.ConnectError, httpx.TimeoutException, HTTP 502/503/504.
• Delay formula: min(base_delay * 2**attempt + random.uniform(0, 0.5), 60).
• Do NOT retry on 4xx client errors (except 429, handled separately).
• Log each retry attempt at DEBUG level with attempt number and delay.

Acceptance Criteria

• A request that fails twice then succeeds is returned successfully after the retries.
• A 400 response is NOT retried — it raises InvalidRequestError immediately.
• Retries respect max_retries; exceeding it raises NetworkError.
• The delay between retries grows with each attempt.
• Retry behaviour is testable by injecting a mock transport.

[Rachaelisa]

Hello, I’ve gone through the requirements and I’m confident I can implement a proper solution that matches the acceptance criteria. Kindly assign this issue to me.