Rather than figuring the "connection" between narrow bars and fragile stems to be any kind of red herring, which to me might imply that maybe someone was trying to be deceiptful(?), I sort of look at it as more of a statistical situation, where frequency of failure was looked at from afar, rolling in the average of a million bikes or so.
This would explain why the industry installed those "death" stems for such a long time, uninterrupted.
The traditional death stems were slowly being phased out on even the half-decent bikes by the mid-1970s, with AVA selling more stems that were looking more "Cinelli" than "death".

Since the industry intitally simply did not sell any relatively wide road bars that were sized for the 25.0mm clamps of the death stems, There was little reason to increase their strength beyond a certain level.

And larger riders who sought wider bars and perhaps longer stem extensions (>10cm) than the death-stem vendors ever produced were forced to look elsewhere (i.e. Italy) for their paired bar and stem combinations, which were available in a much wider range of bar widths and stem lengths.
And these Italian bars/stems would more often also be the choice of serious enthusiasts who were stronger, rode in a more-aggressive position (thus wanting longer stems), and who put much greater mileage on their bikes.
So, no surprise then that these bars/stems (a good number of which might be expected to wind up in the hands of riders who even had brutally-aggressive, track-bike chops) were built to much sturdier dimensions, and of better (usually forged) material.

Now as far as bar width being trendy, one way or the other, it was initially perhaps a bit of an over-reaction to the modest upsizing that was overdue because of the previous failure of the market to size bars commensurate with frame size, similar to the trend toward longer cranks.
Dealers might not complain, knowing their customers might be spending more money later on expensive, wider bars, with the further opportunity to tack the sale of an expensive stem onto the bar sale.

Another thing about bar width has to do with the level of riding that a rider (particularly a racer) is participating in, since the increased stem lengths that developing racers tend to prefer often necessitate a wider bar to more easily control the steering during their most aggressive out-of-saddle efforts (either climbing or sprinting).
That some riders prefer wider bars for their increased leverage in technical or rough conditions creates another group who prefer wide bars for their gravel/cross/commuter bikes, where a wider hand placement on the top/straight section of the handlebar is also much preferred.

I'm mid-sized at 5'9-1/2", and have had to increase bar width on quite a few of the older bikes on which I have installed a longer stem, again mainly to make the bike more controllable during aggressive off-the-saddle efforts.
Typically it is the bikes with a slacker seat tube angle which leave me first wanting the longer stem, and (particularly when these same bikes have a slack head tube angle) a wider handlebar is then needed to keep the steering controllable.
I've had to go as wide as 42cm, and can see wanting to use 44cm for the more extreme cases of long stem and slack steering geometry.

I don't think that aerodynamics really figures into the equation much, except for the original need for installing the longer stem, putting the rider in a more aero and more power-producing stance.
A wind tunnel test might disagree here, but as I understand it, it just depends on the position that the particular rider holds their elbows (in or out). Either way, the wind has to be able to flow between a certain minimum space between the rider's upper arms and their torso, or the rider has to fully tuck their elbows into their gut.