But if the wood does not provide enough resistance for the back to be stretched by, why would it stretch?
It does provide enough resistance, up to the point the belly starts to take set (fail). A belly that has reached this point becomes weaker in terms of stiffness, and then (as as you already know), a hinge happens fast.
When wood is cranked in two by tension, what is the linear increase before breaking? the actual distance?
assuming a 1/2 inch thick limb bent into a 24" radius (at the neutral plane), and the bow is about to break, then the back is at a radius of 24.25", and the belly is at a radius of 23.75".
3.142 x 2 x 23.75 = 149.3
3.142 x 2 x 24.25 = 152.4
if your limb is a 1/4 circle then
152.4/4 = 38.1
149.3 /4= 37.3
or about .80 difference or .4" more than the length of the resting limb.