The "toMarkdown" function prepares content to be sent, primarily, to Diaspora.
The HTML to Markdown converter by default "preserves HTML tags without Markdown equivalents like `<span>` and `<div>.`" At least according to the README in _/friendica/vendor/league/html-to-markdown/_ - which also says "To strip HTML tags that don’t have a Markdown equivalent while preserving the content inside them, set strip_tags..."
Diaspora, however, does not appear to know what to DO with the HTML sent to it. It actually appears to _encode_ the HTML and displays the *code* in the post body rather than rendering it as HTML. In which case it would make more sense to strip out all tags that have no Markdown equivalents.