I believe there has been at least one incident/accident which occured after one of the springs pulled free...
Correct, and your summary of the problem is quite accurate. I remember it very well.... since it happened to me! :shock:
By pure good luck, nothing got bent, and I didn't even know I had a spring failure until applying power to taxi clear of the runway after landing. (Power? What power?) A full description of the "whys and what happened and the fix" can be found in the Canard Pusher, CP65, Oct. 1990 - it's too long to reproduce here. One good approach to preventing this type of problem (spring resonance sawing through an Al bracket) is nicely outlined above.
There is another point that should be mentioned, however. If you rely on a single spring and it breaks, for whatever reason, you could be out of business, depending on the specifics of your installation. (BTW, every one of the four springs on my garage door has broken at least once, some twice, and no vibration involved! - I agree, a lotta ops though.) So anyhow, I originally used two springs in parallel thinking that I had redundancy. But it turned out that neither spring alone was quite strong enough to pull the throttle open on its own (because of the force needed to drive the accelerator pump on my Cont. O-200), so that my 'redundant system' actually turned out to be TWICE as likely to fail! So whatever you decide to do, (I'm currently using a three-spring, damped system), try to be sure that your throttle and mixture controls aren't subject to either single-point or to 'double, single-point failures'. 'Zero, single-point failures' is a lot better plan!

Oh, and you are right about the required springs being extension, not compression springs.