Performance and reliability

Recent versions include several internal improvements designed for large-scale document generation:

• HTML parsing has been rewritten to use Span<char> in critical code paths, reducing allocations and improving parsing throughput by approximatively 25%.

• All remaining regular expressions are executed with explicit timeouts to protect against catastrophic backtracking and potential denial-of-service scenarios when processing untrusted input.

• The overall parsing is done via AngleSharp, and processed in accordance with the Interpreter design pattern.