If Dorico gets what it needs in terms of all the useful key commands in Cubase + all the functionality in Cubase (minus the audio tracks), I wouldn’t need to jump back and forth between the two apps that much. Personally, I haven’t recorded/mixed albums or worked with audio for several years - but there’s still a lot of what you may call “DAW stuff” that’s missing in Dorico. And of course - audio tracks in Dorico has come up a few times in this forum an elsewhere already, but my point is only that adding audio tracks in Dorico (and the relevant functionality that implies) + developing some interchange solution between Cubase and Dorico could represent just as much work as heading towards a unified, ‘modular’ app. And there are dozens of benefits from not having to deal with two apps.
“with the futher development of Play mode, I hope there won’t be any need for midi editing in Cubase. Why not use Dorico for midi tracks and cubase for audio tracks?”
See above (and earlier posts from others than me) about the benefits of having one app to deal with. And - to use myself as an example again - Cubase and Dorico has some clear benefits over Logic, which has been my main tool ‘forever’. If I would have to learn one new tool, and not two (of which one of them - Dorico - still is missing some of the stuff I find essential (eg some Sibelius stuff), switching would have been a lot easier.
“True, but then we would also need a “Cubase with or without pro audio functionallity”, or it would be too expensive for those only interested in Dorico’s engraving capabilities.”
I don’t think finding a way to enable different parts of a program is an impossible task It’s of course more complicated than just making Cubase or just making Dorico. But is it a lot more complicated than making both? Even if the answer would be yes, I’m convinced that it would be a solution that’s much better for almost all users.
“The model of music notation in Dorico is completely different to the data model in Cubase”
Well, I have no idea what it would take for Steinberg to offer a solution which would - for the user - appear more or less as if one would have to deal with only one app. I’m just saying that such a solution would make diving into the Steinberg world a lot more tempting, and I know that many Steinberg users also really would like such a solution. It would serve as a strong argument for going Steinberg for all current and future Logic+Dorico users (and many other Dorico users who don’t use Cubase).
“I just meant that zoom, transport, and other common functions should be the same (I think some of them are already). Uncommon functions should be optimized for each app.”
Well, I know this is how many Logic users see it: they/we use a dedicated score app not because the Logic score editor is horrible, many of us actually think it would be tons better if it only would have kept being developed by Apple. There’s even stuff in Logic’s score editor that some of us find superior to what the dedicated score apps offer. Many of us would never consider any of the the three main score apps if Logic only would have, say, 17,5% 
more (silly with a percentage here, but forget that for now) more score functionality than it has.
So if Apple can make a combo app that’s so close to being what most Logic users need, why can’t Steinberg make a combo app that also only lacks those silly 17.5% I mentioned - but spend a handful of years to add those missing things? Steinberg is serious about notation, they don’t suffer from the pop culture thing Apple has gone into, they are serious about Kontakt, with working with Kontakt libraries etc… Again: Logic is closer than Steinberg in terms of being capable of delivering such a product (and who knows, maybe they are working on it), but Steinberg is closer in terms of having the needed people, knowledge and (drum roll…) interest in the stuff I’m talking about here.
"but my point is, I don’t feel too much compassion for those saying “I want Cubico, because it’s so hard to learn KCs for two applications”
The KC part of this is only a little part of it. A product - or merging two products (or pseudo-merging them - is about much more than key commands. Most of all it’s about providing a workflow that doesn’t need loads of interruptions because you need to do some of your important tasks in one app, and other tasks in another. Editing music with decent looking notation and using the many advanced MIDI features the best DAWs have shouldn’t have ti require two apps.