Well, increasing the resolution of the sequencer could be achieved in at least two ways I can think of now, but one of them (the more promising one) is something I would really need to build a experiment for. The other one bases on the current model and seems very CPU heavy, but still worth validating.
What needs to be rewritten in the way the sequencer events are generated and put into the sequencer queue, and the way the sequencer queue is then played back.
There is a component of latency that I need to investigate an experiment. The OS values in the specs are good, but I need to see it running, and hear it play. Realistically, that looks more like a summer project rather than a quick fix.
Ok, some years the summer comes earlier, especially in these days.. so will keep you posted on any progress / realization I manage to achieve at that front.