It even uses the exact same syntax as in html, so as long as svg content isn't specifically excluded, normal web scraping would just work without modification.

Perhaps, but I think OCR is more likely.