Just creating an issue to keep track of this outstanding to-do from #454: we should add a way of encoding the visual style of a multi-note tremolo.
The primary case for this is tremolos in which the displayed noteheads are half notes. Some publishers prefer these to be beamed with a single beam (while other beams are disconnected), others use all beams connected, others use no beams connected.
We could provide a way to encode this decision, both on a per-multi-tremolo basis and at some global level. I didn't implement it as part of the #454 changes because I wanted to wait until we had a way to encode global styles.