I have been noticing this in plate engravings too lately. Sometimes I ignore the difference and let Dorico draw the hairpins its (logical) way; sometimes I try to attach the endpoint at a beat in between the notes that gives a similar appearance; and sometimes I’ll actually end the hairpin one note earlier than it would appear to function.
These 3 different approaches correspond to the 3 samples in the OP. I wouldn’t bother accommodating the tremolo example because surely you want the crescendo to continue through the last note. The Liszt example does need to be reproduced faithfully. And for the octaves example one note away is close enough.
Overall I think hairpin positioning needs to be treated case-by-case almost as often as immediate dynamics. An entrance on a note way above or below the staff often wants the dynamic to the left of the note to save vertical space, etc.
If you are on a quest for uniformity, that runs into the fact that hairpins have many different meanings depending on context, and each deserving its own graphical representation.
If you are looking for authenticity, that runs into the fact that composers tend to be sloppy about positioning hairpins and engravers have just guessed at the intention, sometimes imitating slavishly with absurd results and sometimes according to their own style.
If you are looking for a more logical system for hairpins, you should consult first editions and manuscripts when possible, analyze what each hairpin actually means and devise a system that fits the various situations that arise. And Dorico should allow you to make use of all of these possibilities.
The Liszt example points out that dynamic placement for a piano work might be very different if (say) the half notes in the top voice had been played by a violin. Trying to program dynamic end-points differently for C1/C11 instruments vs. velocity-controlled instruments would be an enormous challenge, so I expect some manual adjustments to always be necessary in some cases.
That said, multiple options for how to handle dynamics’ end-points (as is possible in one case already) might serve a useful purpose for some writers/engravers.
In the first two examples, the hairpins are tilted, even though there seems to be enough room to align them horizontally. But - the tilt seems to be following the contour of the notes. Am I imagining something, or is this a valid technique (tilting) for drawing hairpins in certain instances? Or just a bit of sloppiness on the part of the engraver?
Tilting is a perfectly valid technique, but these days it is usually used as a last resort. In earlier times, hand engravers seem to have been desirous of having beams and hairpins follow the contour of the melody in its entirely and less concerned about following rules that focus entirely on small groups of notes. As a result, to me their engraving looks more human and musical than much of what we see today. I think we could learn a lot from the old hand engravers.