Looking at the flutes, I’m not convinced this is anything to do with grouping dynamics.
Dorico splits the music into “phrases” and independently chooses how to condense each phrase. To simplfy what it says in the release notes, phrases are separated by rests, unless some other considerations extend them over a rest.
In the first phrase you have different note values (half notes against a whole note) so the condensed part has to use two voices, and they are used for the complete 4-bar phrase.
In bars 12-13 there is nothing to stop the all the notes being in one voice.
In 28-30, again you have different note lengths.
I’m not sure about 39-42, but I think the tied whole notes will count as a single note, longer than the two whole notes in the other flute.
In bars 100 on, the unison whole notes probably make Dorico choose two voices (I’ve noticed in my own scores that “a 2” never seems to occur on whole notes for some reason). Because of the simplistic idea of a “phrase”, the condensing is all done the same way between 100 and 124 because there are no rests.
The same ideas seem to explain what happens in the trumpets and trombones.
Published orchestra scores usually avoid frequent switches between two stems and one, unlike a keyboard part where you would normally use two voices on a staff for the minimum possible number of notes.
I think the real issue here is that until there is some user control over what is a “phrase” in a future version of Dorico, some things are inevitably going to be sub-optimal - but your score is perfectly legible, even if it’s not as beautiful as it might be.
FWIW my biggest complaint about the current version is that it is over-enthusiastic about condensing parts onto single stems, with lots of “a 2” indications on isolated notes. Using two voices for the whole passage would look better IMO. In published scores, “a 2” is normally used only for unison passages that continue for several bars, not for single notes.
