The problem with using springs is they store energy, and then release it.
A spring delays the effect from one end to another; in both directions.
For example; a car when stationary doesn't weigh any less because you use a looser spring; the weight bearing down on the ground is the same whether the spring is stiff or not.
The variable is
time. The spring will 'absorb' impact when you drive, because each jolt will be short duration. If you give the spring enough duration (when stopped), your car will be applying an equal amount of force to the ground regardless of how stiff your spring is.
Its not that the energy is "lost", its that the feel is lost. Its like f@^king with a condom. The promlem in this case is also that the delay in energy application negatively effects the ability to break bolts loose. This is just empirical fact, regardless of equations.
Empirical fact? That's not a thing.
The "delay in energy" does not negatively affect the ability to break bolts loose. The force on both ends is the same over time. When we talk about very high spring rates like what we are talking about here, that delay in time is extremely short (milliseconds).
Its common for people to assume they need to apply more force when they see torsion. Which causes that snap. Just go slower, its a duration thing, same result.
If you spread out a force over time (think hammer blow) and you lose some useful properties there (hammer= concentrated force in small time). When breaking free a rusty bolt, you want to deliver force in quick-time...and thus you want to use use maximal stiffness. IE, maximally-sized square drives and maximal extensions for superior .
When using an impact wrench yes; primarily because the 'hammering' effect does not last long enough before it is released and the next hammering effect gets applied.
When using hand tools, there wont be a difference, unless you are hammering your ratchet.
You can put a SO 1/4 drive ratchet x 10 inch on 50 ft pounds, but it won't break bolts as easily as using 1/2 drive x 10 inch. The 1/4 drive will start to look like a pole vault at some stage, and while it will devliver the torque (probably in some experiment) applied with a torque wrench, it won't deliver a predictable, maximal torque. Because the definition of bending is the torque to bend is lower threshold than the torque to turn bolts. Therefore, what you feel and what is applied is disconnected, and this creats crappy feedback to the end user.
This whole part is just down right incorrect.
Newtons third law doesn't care if you are using a 1/4 ratchet or a 1/2 ratchet.
"Feedback" or feel is a matter of time, go slower, you will feel just fine.
Crappy feedback also leads to people pulling too hard, and putting excess energy into the sytem. If you are expecting 100 ft pounds is needed, and you push a spring until it it takes 100 ft pounds, there is alot more energy into the spring than if the spring was a non-spring. As with suspension, the purpose/function of the spring is not not allow the 100 ft pounds of that pothole hit you in the ***.
I agree that people sometimes feel the need to over apply force because they misunderstand what is happening.
However, as I noted above, the purpose of the spring is to
delay that 100 ft pounds. If you apply 100 lb to a spring and wait, you will get 100 lb on the other end.
The point about needing to apply more than 100lb into a spring to get more than 100lb on the other end is just down right incorrect.
This is why you don't put springs in between your ratchet and the bolt. YOu want the ***-blasting force to go to the bolt.

This is why you have transmission extensions on 1/2 square drive female with 3/8 male-drive for the smaller sockets. Some mix of oversize dimensions in-between for more stiffness with appropriate clearnce for the job. (etc).
Again, only applies if you are talking about an impact wrench. For a ratchet, it doesn't matter how soft your extensions are, the force will always be the same on the other end given time (milliseconds).