Alright, without getting too technical, the process by which the "reset" the spring is essentially called that as it's meant to be the equivalent of starting again with the spring, just achieving a shorter length with the same properties, but this doesn't happen. When the steel is heated above 1000 degree they're returning it to the Austenitic phase, which would normally decompose slowly to ferrite and cementite in equilibrium conditions (don't worry too much about the names, just know that's what it turns to). The steel will have very few microalloying elements in it, and as such a high martensitic start temperature. As the springs are quenched in an oil bath (better than water, but still not equilibrium conditions) you'll get thermal stresses created within the springs whereby the exterior cools quicker than the core. This will cause the interior of the spring to be exerting a constant force on the exterior, which will also have a higher martensitic content. When i say martensite, read really hard, but really brittle. A little bit of martensite = good, a lot = bad.
Ok, so you've now got this constant stress on the inside of the spring that puts it close to its elastic limit. The springs are now work hardened, but attempting to undergo the same application as the original springs. With a work hardened spring, it takes a smaller elongation (or compression) to reach the elastic limit, but at a higher stress. Although this higher stress gives you a slight pillow to the property change occuring within the spring, the overall fracture toughness (energy able to be absorbed by the spring is lower). As such, one sharp bump from a pothole or missed speedbump or whatever is more likely to fast fracture the spring.
The other problem is the cyclic fatigue of the spring will be deteriorated. Normally a spring would be mate to operate in a region whereby cyclic fracture would not be seen to occur, by work hardening the spring, you've increased the mean stress the spring is seen to undertake, putting it into the cylic fatigue region. This gives you a certain number of cycles (read 10^5 to10^7 cycles) before the spring will be seen fail.
You'll also notice people talk about the springs sagging massively, this is due to the cyclic loading weakening the springs within the plastic region (elastic = recoverable damage, plastic = damage that isn't reversed upon unloading), basically as a slow prelude to their failure.
Ok, so that was a bit of an epic post
, but, from a materials engineering standpoint, that's why you don't want to reset/recompress your springs, and shows that we're not just being "haters" when we say it's not the best to get it done, there are reasons behind it! But it's up to you mate. Any questions, feel free to ask.