the problem here i think doesnt lie with the numbers of enemies but peoples reluctance to train certain patterns like evading dogs, nets or getting a mind-map of enemy priority like trapper>dog>flamer>bomber>gunner>horde trash etc.
you see people fail on malice to a sole trapper, each… one after the other…
i´m sure you´ve seen a neat railing from which were hanging 4 ripe rejects…
you get the idea 
sometimes there´s just a gap people can´t or wont bridge for one reason or the other.
maybe as a developer you should accept that not everyone is going through all the difficulties as a “progress” and willingly or unwillingly stagnates at some point, but still make the bar flexibly higher for those that do.
playing auric maelstrom or HISTG no matter if tired or as a warmup i see my path up until here as a natural progress i´d like to continue.
last couple weeks i broke my latest plateau that was building up karsolas like speed with my branx and combine the best of two worlds.
now i can traverse as easily with my oneshot machine as i did with the lighter version 2-3 months ago.
practise and dedication, still not there with the auric maelstrom solo-crowd but i want to challenge myself in every way possible the game provides.
as for the AI ramp up, i totally agree, though i got my doubts the games engine will allow for much given the current trapper shoots through a phalanx of bulwarks “collision”
i´d love for groups of gunners to retreat, try a different path, flank or call other units over for example but seeing how ai stagnated since half life i wonder if it´s an engine/hardware thing.
game needs to sell on lower end hardware too after all.
but maybe add another pattern to the trapper as to not shoot a horizontal projectile but lop it at you at an arc and catching on from above.
that way you could give the trapper a new movement set as well.
i´m no game designer but i think you get it 