First, I assume the spring is guided on a shaft or a hole and if the inside or outside diameter of the spring touches the shaft or hole there may be a friction between the inner/outer side of the spring and the shaft/hole and the 66.388 in-lbs is composed of the spring resistance and the friction with the shaft/hole combined. If this is the case then the spring may see much less torque than 66.388 in-lbs.
The minimum ultimate tensile strength of 0.135" Music wire diameter is 258ksi however, the maximum is 285ksi. Therefore, it may be that the actual wire you are currently using is on the upper level of the spec and therefore you do not see any detectable set. However, you have to design for the minimum spec values. There is also a chance that the actual wire diameter is more than 0.135" (in the upper level of the wire diameter tolerance).
According to the Associated springs Company (SPEC) for static applications you can design for up to 100% of the ultimate tensile strength “for springs without stress relieve” and 80% for stress relieved springs. However, your case is definitely cyclic. The SMI Handbook of Spring Design does not discuss the stress levels issue.
In each cycle your spring approaches UTS therefore, from fatigue design point of view (on the SN curve) the spring life should be less than 1000 (out of the graph). The reasons I mentioned earlier may be the cause that your spring in the tests show higher life cycles.
You should notice that the spring inside diameter may decrease to 0.952 inch when maximum loaded.
Can you provide the guide shaft (hole) diameter?
Can you describe how the spring is loaded? Is it loaded with pure torque or it bends due to the force on the arms?
How many of the 241 coils are truely active inactive?
How long are the arms?