I had a similiar idea. But to do it in opposite manner. To synchronise visuals to the music. Of course music would be generated on the fly, tracker way, this would give the most control over particular sounds. And when you posted this message, I started to think, how it would fit into elevators. Imagine all (at least all empty) elevators moving in sync with music.
Very cool idea, like many games that sync to the music but a step further. There are definitely a LOT of interesting doors that can open using tracker-based music. For the record, the video above was running OpenMPT with a bunch of VST plugins, while recording in OBS, and still hitting 90fps on my 6-year-old Core i7... trust me, it struggles with a lot of games, so the fact that it could do all that without a hitch is a testament to your design... very efficient! I could easily rework the music to ditch the VST plugins for compatibility with general IT playback plugins, which would even further reduce CPU overhead to negligible.
Now there's an idea. A section with platforms, so you can see a lot where enemies dance in sync and they shoot in sync with the beat. I have no idea if I will have time to do that, but I will have to try it out. For time being I plan to work on pure gameplay features alone to get to proper vertical slice. Then I might work on music system a bit.
The game is using multithreading a lot. Because of that it can hit 90fps. The bigger the level, the more workload there is - all NPCs moving around, etc, but there is a lot of cheap optimisations, ticking some stuff every second or third frame etc. With 90 fps some enemies have logic running at 30fps. Which is what is happening for some of the normal games ;)