Add support for Markdown conversion

In ingestions coming from OpenReview (e.g. #8710), Markdown syntax is often found in abstracts as OpenReview supports this. Our library (and therefore ingestion pipeline) currently only performs LaTeX conversion. We should consider if some basic Markdown can be supported as well, i.e. adding `MarkupText.from_markdown()`.

In practice, we almost always want to convert heuristically – i.e. we don’t know in advance if an abstract we ingest contains Markdown or LaTeX. This will probably require careful testing to ensure nothing breaks by accidentally interpreting Markdown as LaTeX or vice versa. The current function `MarkupText.from_latex_maybe()` should probably just be replaced with something that performs all these heuristics, e.g. `MarkupText.from_any_markup()` or something like that.