I think Nikola has summed it up well - software development is hard. The thing that may not be apparent is that unlike a sequencer, Dorico doesn’t actually store any MIDI at all. What you see in the Play Mode has to be generated from the notation and then any edits have to keep the MIDI-like and notation representation in sync. So there’s a huge amount of work that has to go into making anything in Play Mode visible or editable. It’s not simple to let you have more control, though we really want to greatly increase the amount you can do in Play Mode over time. It will never be a proper DAW, but we expect it will borrow from the DAW functionality you expect, where that makes sense (creating notes, drag/drop, automation lanes, etc).
Building a notation layer over the top of a DAW wasn’t an option because first and foremost Dorico is for music notation, and that is the representation that is the 1st order citizen in our data model. However, the audio engine is basically a very stripped down version of Cubase, and effectively Dorico constructs tracks of MIDI data as you would in Cubase, and feeds that to the engine. However, there is a lot of complexity in that translation layer.