Gitto - re your #16 and #21 -
I think you're being inconsistent here, complaining about the use of characters, other than straight English Latin alphabetic characters, "...that cannot be resolved within the [Listener] crossword itself" Surely we've had examples using scatterings of Greek, Cyrillic, and Hebrew characters in the past - even non-letters - none of which could be deduced without outside research, except by the most linguistically polymathic of solvers; and how about thematic concepts such as, for instance at random, the Brunel Box Tunnel sunrise, and the Flann O'Brien 'At Swim Two Birds' burning hotel, which are essential to the final solution of most Listeners, and which couldn't possibly be resolved by deduction from "within the crossword itself".
'Outside' research is obviously - to me at any rate - essential.
The giant brains that not only can solve the clues, but also have all the onboard information that enables them to deduce the details of every theme, however arcane, without reaching outside the grid, must be lonely indeed.
How people managed with only the BRB, maybe a shelf-load of Britannica, and the ODQ, before the availability of powerful encyclopaedic and analytic online devices like Google, Wikipedia, Chambers App, and anagram solvers - and Bradfords, I can't really imagine.
(...although on consideration, when I started attempting to solve Listeners many decades ago, I must have been doing exactly that!)