There is some vestigial code when importing HTML files which first converts the HTML to Markdown and then converts that to ProseMirror.

With two conversions, this is error-prone and unperformant. Now that Markdown is no longer the canonical storage format, this is also unnecessary.