JoachimM, to address you question, "Why would a bottom limb have to be strained differently from a top limb, especially in symmetrical bows?"
That's an easy one. In order for the limbs to share the load equally, one limb will need to be made weaker than the other because we hold a symmetrical bow at the center, but pull the string from almost 2" above that(shooting split finger). 'Simply' put, pulling the string above center means the distance from the string fulcrum to the bottom tip is longer than the string fulcrum to the top tip. This makes the portion of the string below the string hand a longer virtual lever than the upper, giving the LOWER portion of the bow a mechanical advantage. This advantage means it needs less strength to do equal work(flex) with the same load, i.e. it should be tillered to show weaker at brace height (very generally speaking of course) in order for the bow to be balanced at full draw.
Staying with symmetrical, even-tillered bows for this example... Moving the string hand above or below center on the string shifts the advantage. It's immediate and simple to feel... and obvious to see what it means/causes as a result. The slightest tug on the string tips the bow. Move the string hand up, we give the bottom limb the advantage and the top limb tips towards us.... move the string hand down below center, and the top limb is given the advantage, causing the bottom limb to tip towards us. The way to balance the bow relative to this, to cause equal amount of flexing in the limbs, is to weaken the limb with the advantage.
"I always aim for a perfectly symmetrical tiller and don't leave the lower limb stiffer."
I don't leave the bottom limb stiffer either. I balance limb strength at full draw relative to how the archer will hold and pull, then the tiller measurements at brace height are whatever they end up being... always a result, never a predetermined guide/guess.
You said, "Now, if you leave one limb stiffer ("not weak enough"), it means the other (necessarily weaker) limb takes more than half of its share of the stress, and is strained more. More strain, more set. So shouldn't then the weaker (upper) limb take more set?"
Yes, IF we pulled the string from center, but we DON'T. Since in reality we pull the string away from center, the answer is NO. Not necessarily. Your statement ignores the physics of levers mentioned above. In addition to that, stiffer as viewed how? At brace? What does equal(or ANY) tiller at brace mean when one limb has a mechanical advantage over the other?
The strain the limbs ultimately realize is due infinitely more to where and how we grip the bow, where and how we grip the string, and the strength of the limbs relative to one another WHILE those factors are in play at full draw, than how they look compared to each other at brace... static... yet folks insist on using the profile at brace as their unequivocal guiding light to gauge/assume relative limb strength. That can be a critical mistake, in my opinion.
That braced profile so many sink their trust in, is practically meaningless to me during construction. After the bow is finished being tillered on the tree and immediately prior to taking it out to shoot, I take the tiller measurements for the first time... just so I'll know if the tiller changes during shooting in... but it virtually never does... because, ah-ha, I tiller the limbs to see equal 'relative' strain.
By the way, since I do, my bows don't take more set in the bottom limb. If/when they take set, they do so evenly along the bow's length. I could show you dozens of examples. As a result, never once either have I had to flip a bow end for end because the tiller shifted.
It's things like this, that cause me to question the degree of relevance of m.c. differences between top and bottom limbs... in other words, I FIRST wonder how such bows were balanced relative to the shooter at full draw.