You get to attribute p (high kinetic load) to the first floor(s) of falling building, but the moment you have impact, the v of that lower floor is roughly zero, as it is simply crushed against itself.
The highlighted part is misguiding. At the very instant of contact, it's at zero velocity, but as the impacting floor pushes on it due to its inertia, it gains velocity *very* quickly, basically instantly. Maybe it will help you if you think about it in terms of a bullet hitting a block of wood: the block of wood gains velocity very quickly too when impacted, as the bullet's inertia pushes it. The resulting theoretical velocity can be calculated by the conservation of momentum formulas.
That velocity is unlikely to be reached(*) because there are several energy sinks, one of which is the energy inverted in breaking the connections of the floors to the columns. But still, given that these connections are designed to only support the static load of one floor (times a factor of safety) and not such a dynamic load, the energy loss due to that alone will be quite small, so I don't expect that to be an important factor.
(*) It can be reached for other reasons. A partially elastic collision may make the elastic part compensate the loss of energy and make the bottom floor reach the same theoretical velocity as in a perfectly inelastic collision. But that's nitpicking. Almost surely that didn't happen.
To summarize, the final velocity of the impacted floor an instant after impact will not be zero, so it will have an initial KE (which depends on the initial energy of the impacting floor) that gets added to its initial PE, both available to keep crushing, AND the impacting floor will still have some of its KE available and the rest of its PE.
The added PE is nowhere near the PE you got when x number of floors got to descend through y vertical space.
Actually, my initial assumption was that one floor alone is able to break the connections by impacting on the next floor, making your x = 1. Under that assumption, the impacted floor
alone has enough energy to continue the process, even more if we consider that the impacting floor still has PE.
But let's take the case of x = 5 floors as the minimum number required to fall for the height of 1 floor and break the connections. Under the assumption that the loss of mass due to lateral ejection is about 20% of the mass of one floor on each impact, we have the following situation: the total mass in movement after the first impact is now 6 floors minus 20% of one floor = 5.8 floors, falling for the height of 1 floor,
regardless of any loss of energy (we can assume they start again with 0 velocity, in
Judy Wood's Billiard Ball model's fashion). Since this floor didn't resist 5 floors falling on it for the height of one floor, there's no way the next floor will resist 5.8 floors falling on it for the same height. Much less if they have an initial velocity, as will be the case.
In practice, x is probably 1, and almost surely not bigger than 2. Again, these connections are not designed for such a dynamic load as that of one floor falling on it.
So you get less KE after impact, less to no PE released, and a system which comes to a halt well above the ground.
The highlighted is false. The impacted floor will be detached, and since it's no longer held, its PE will start converting into KE. Meanwhile, the impacting floor has about the same PE as the impacted one. Actually more if it's a block rather than a single floor.
The conclusion that the system comes to a halt, is wrong for all the reasons above.
I'm not sure what your definition is, but if it departs from the reality of the towers' collapse, it's not useful for discussion purposes.
No column crushing happens at this stage. The KE loss you refer to is inherent to inelastic impacts. Once again, your assessment applies to the horizontal case, but gravity adds the ingredient that allows the collapse to continue all the way to the bottom.
I suggest you to google "progressive collapse" or "disproportionate collapse" to learn on how the real world structures behave. Same for gerrycan.