As I am sure Susquatch and others know already, there is a lot of stuff being skipped in this analysis.
Assuming elastic behaviour only
Every force in compression produces forces in tension at an angle. Generally, around 45 degrees. Think about using a tractor to grade a hill of dry sand. If it gets any steeper, the sand just slides sideways.
When a load is applied to a solid object, it spreads out in the same kind of way. Up down or sidesways, the load makes 'pyramids' from the source to the resistance. The exact shapes vary dramatically and depend to a great degree on the contact surfaces involved. Usually they are complex shapes where any sharp corner represents a stress concentration - AKA a stress riser - and a weak point
Absolutely.
There is a lot more that it relevant about apparent load, moment length tributary area and failure modes, but I don't think I have ever considered any of these factors when bolting anything to my table
You obviously understand way better than most.
The three problems that raise their ugly head in this particular situation are:
1. The compressive strength of cast iron is about 5x its ultimate tensile strength.
2. Most steels have ultimate tensile strengths that are only slightly higher than their compressive strength.
2. There is this nasty stress riser on the inside upper outer corners of the Table T-slot.
By design, bolt threads are capable of developing extremely high tensile or compressive forces.
That is why most T-Nuts are not threaded all the way through. This prevents users from threading a long bolt through the nut and then loading the T-Slots to failure.
But the same thing can happen with T-slot bolts even if they don't reach to the bottom of the T-Slot. If the setup is such that the steel can deform within its elastic range but above the failure limit of the cast iron T-Slots, they can and do fail.
But even if I didn't know this, I have seen enough broken T-Slots in my time to know that it's a lot more common than most people realize. Here is a random screen shot of a Google search for "broken T-slot". Easy to see why I call it tear out. Very common.
My guess is that your knowledge of stress and strain has been your friend. Basically, you know that it doesn't take much torque on the bolts to develop extremely high friction between the parts and the table. So you don't over-tightened them and therefore you have never had a problem. That doesn't work for everyone. Far too many people over tighten those bolts and put their tables at risk.
Again, I've seen it far too many times and the reasons are always the same - a T-slot excessively loaded in tension.
That's why some guys say never buy threaded through T-Nuts and why I advocate only using T-Nuts and hold downs in compression. It's almost impossible to overload a clamp to the point where it can tear out a T-slot if it is fastened in compression.
I can appreciate that your experiences have been good for the reasons I suspected. If you can, I'd suggest you see if you can get your hands on a scrapped table. Then do some experiments. I think you will be surprised at how easy it is to pop a T-slot in tension and how impossible it is to pop one in compression - in most cases, the bolt will strip first simply because of the area difference.