feat(utils): add retry utilities with exponential backoff

[jmagly]

Summary

This PR adds retry utilities to utils.py, addressing E2.6 from the roadmap ("Add retry logic for network failures").

What's Included

Two new functions in src/skill_seekers/cli/utils.py:

1. retry_with_backoff() - Sync version for requests operations
2. retry_with_backoff_async() - Async version for httpx.AsyncClient operations

Features

• Configurable max attempts (default: 3)
• Exponential backoff with configurable base delay (doubles each retry)
• Operation name parameter for meaningful log messages
• Proper exception handling (re-raises last exception if all retries fail)

Roadmap Alignment

• E2.6: Add retry logic for network failures ✅

Usage Example

from skill_seekers.cli.utils import retry_with_backoff

def fetch_page():
    response = requests.get(url, timeout=30)
    response.raise_for_status()
    return response.text

content = retry_with_backoff(
    fetch_page,
    max_attempts=3,
    base_delay=1.0,
    operation_name=f"fetch {url}"
)

Async version:

from skill_seekers.cli.utils import retry_with_backoff_async

async def fetch_page():
    response = await client.get(url, timeout=30.0)
    response.raise_for_status()
    return response.text

content = await retry_with_backoff_async(
    fetch_page,
    operation_name=f"fetch {url}"
)

Log Output Example

WARNING: fetch https://example.com failed (attempt 1/3), retrying in 1.0s: Connection refused
WARNING: fetch https://example.com failed (attempt 2/3), retrying in 2.0s: Connection refused
ERROR: fetch https://example.com failed after 3 attempts: Connection refused

Tests Included

7 new unit tests in TestRetryWithBackoff and TestRetryWithBackoffAsync:

• test_successful_operation_first_try
• test_successful_operation_after_retry
• test_all_retries_fail
• test_exponential_backoff_timing
• test_async_successful_operation
• test_async_retry_then_success
• test_async_all_retries_fail

Integration Note

This PR provides the utility functions but does not modify existing scrapers. Once merged, the doc_scraper can be updated to use these utilities:

# In doc_scraper.py scrape_page():
def _fetch():
    response = requests.get(url, headers=headers, timeout=30)
    response.raise_for_status()
    return response

response = retry_with_backoff(_fetch, operation_name=f"fetch {url}")

This keeps the PR focused and reviewable. Integration can be done in a follow-up PR if desired.

---

Contributed by the AI Writing Guide project.