Comment by rcxdude

Nothing prevents you from attaching your async code to a high-priority interrupt (you will need to be careful about synchronization, but no more so than in a regular RTOS). You can 100% do this kind of thing, and you'll probably get better latencies as well.

(interrupt queuing and prioritization can be handled by the hardware nowadays. RTIC, a similar project, uses this to drop the need for a scheduler altogether and just runs all code in interrupt handlers, something which 'traditional' realtime wisdom would forbid because it's from an era where you didn't have something like an NVIC.)