Retry mechanism for transient errors

[Mews]

Closes #39

Changes

• Added a is_transient_error function inside fetcher.py;
  • I hardcoded the list of transient errors as [408, 502, 503, 504] since they're the most common ones, but let me know if I should add/remove any;
• Modified fetch_url to retry fetching the url when it encounters a transient error until retries reaches 0;
• Added tests to test the retry feature both directly through the fetch_url function and in the crawl method;
• Added a max_retry_attempts option to CrawlSettings;

Ps.: The wait times for each consecutive retry attempt are just 1, 2 ,3, ... seconds. Let me know if that's ok.