Fix thread safety, resource leaks, and error handling across core and external modules

[dpol1]

Summary

This PR addresses a set of code quality, correctness, and robustness issues found across the core and external modules.

Thread Safety

• Replace non-thread-safe SimpleDateFormat with DateTimeFormatter in FileResponse and DateUtils.parseDate() in CookieConverter
• Replace shared mutable Matcher in RefreshTag with a Pattern constant and per-call Matcher instances

Resource Leaks

• Use try-with-resources for FileInputStream in FileResponse and InputStream in RegexURLFilterBase
• Add cleanup() to DebugParseFilter to close the output stream
• Ensure Playwright tracing is always stopped via a finally block in HttpProtocol, even when an exception is thrown

Error Handling & Robustness

• Wrap Long.parseLong() / Integer.parseInt() in FetcherBolt with try-catch to handle invalid metadata values gracefully
• Add null check for classpath resource stream in RegexURLFilterBase
• Fix duplicate setExpiryDate() call and add null guard in CookieConverter
• Fix misleading error message in CloudSearchUtils ("must be score" → "must NOT be score")

Logging & Code Cleanup

• Replace all e.printStackTrace() calls with proper SLF4J LOG.error()
• Replace URLEncoder.encode(url, "UTF-8") (unnecessary checked exception) with the StandardCharsets.UTF_8 overload
• Replace manual MessageDigest boilerplate with DigestUtils.sha512Hex()
• Use actual document charset in JsRenderingDetector instead of hardcoded UTF-8

No behavior changes intended

All changes are backward-compatible. The crawl logic and output remain the same.

[jnioche]

Thanks @dpol1. Having a PR affecting loads of different files and covering such different aspects does not make it easy to review. Instead of a large dump, having one PR per type of issue (e.g. Resource Leaks) means that changes can be reviewed quicker, merged in isolation and also reviewed in parallel.
Any chance you could break it down a bit?

[dpol1]

Sure, sorry about that — got a bit carried away. Will split it into three PRs: thread-safety, resource leaks, and fault tolerance/cleanup.