Sitemap generation encoding bug

We're encountering a perplexing issue with our 'Free XML Sitemap Generator' following a recent update. Specifically, we're seeing inconsistent character encoding for non-ASCII URLs and content within the generated XML sitemaps, which subsequently leads to validation errors when external tools attempt to parse them.

Our system is configured to use UTF-8 consistently throughout the entire sitemap generation process. However, when URLs containing characters like 'é', 'ö', or 'ñ' are processed, external validators often report them as malformed or substitute them with the � (replacement character). What's particularly puzzling is that our internal XML parser correctly interprets these characters without any issues. For instance, a URL like https://example.com/produits/électronique might be generated correctly by us but then flagged externally. Here's a conceptual snippet of what an external validator might report:

ERROR: Invalid character (�) found in URL.
  Line: 15
  Column: 87
  Context: <loc>https://example.com/produits/�lectronique</loc>

This suggests a deeper character set handling problem during the final XML serialization or perhaps how certain validators interpret UTF-8 byte sequences. Has anyone dealt with deep-seated character set issues during sitemap generation, especially when interacting with various XML parsers or third-party SEO validation tools? Anyone faced this before?