I think you can totally expect better effects, probably with focus on "less complicated" and "latest DSP".
Drum Kits may happen, but probably different than one would expect. We're trying to incorporate the research we've done on physically modelled drums, but we have no final opinion yet on how useful this will actually be. Furthermore, instead of multitimbrality we would go for some kind of sub patch selection, like "this voice plays these modules, and those modules here are silent". We have yet to check and see if this ids a viable concept.
Similarly, I'm not such a big fan of replicating DAW business in soft synths, hence I do not think a piano roll is what I'd do. I did however learn a lot about modular sequencing in recent years. Zebra 3 will certainly have polyphonic and polyrhythmic sequencing options along with quantising and maybe that above mentioned timbrality control.
The obscure concept of a hybrid arpeggiator / sequencer is a thing of the past, the combo of independent arpeggiator and sequencer like in Hive is just more flexible. If we have the nerve, we'll explore polyphony in this as well, like Arp+Seq per MIDI channel or something. No promise though, this can easily become a dead end during development.