feat(core): update request retry statistics in real-time

[harryautomazione]

Description

This PR resolves the issue where the requestsRetries counter and the requestRetryHistogram were only updated once a request was fully handled (either succeeded or failed permanently).

Updating these statistics only at the very end of a job lifecycle created a false impression of no errors during long-running crawls, and led to incomplete statistics if the crawler crashed or was migrated early.

Changes

• Core/Statistics:
  • Added registerRetry(retryCount: number) in the Statistics class to dynamically increment the global requestsRetries and update requestRetryHistogram in real-time as soon as a retry is triggered.
  • Added input validation guards inside registerRetry and _saveRetryCountForJob to prevent NaN or negative values from corrupting the statistics.
  • Implemented automatic array backfilling (0) inside the histogram to avoid sparse arrays which serialize to null in JSON.
  • Refactored _saveRetryCountForJob to only record job completions at index 0 if no retries occurred, preventing duplicate counters.
• Basic Crawler:
  • Hooked this.stats.registerRetry(request.retryCount) within the _requestFunctionErrorHandler right after updating the request's retry count.
• Tests:
  • Added comprehensive unit tests in statistics.test.ts to cover real-time tracking, negative/NaN input protection, and JSON serialization integrity.

Fixes #2732