Two things.
The role of the tailplane is to provide both stability and control. Even if the aircraft were stable tail-off, one would still need a pitch control surface....
And the percentage of lift argument is, I'm sorry, nonsense.
Think of this case:
Aircraft with a wing, tailplane, and reaction jet at the end of a long pole - not unlike the pitch 'puffer' on Harrier.
Case 1: pitch puffer off, tailplane generating 0.4 CLt to trim the aircraft. Change of alpha at the wing creates a change in alpha at the tail of 1 degree, which gives a delta CLt of 0.1 and a delta pitching moment of 0.1*k
Case 2: pitch puffer ON, tailplane now only generates 0.2CLt to trim because the 'puffer' is doing the rest of the work. Same change in alpha and alpha tail, same delta CLt of 0.1, same delta pitching moment of 0.1*k
What matters for pitch stability is the change in pitching moment with change to aircraft alpha (or aircraft lift if you prefer - the classical textbook way of showing stability is the dCm/dCL plot). In both of those cases the tailplane contribution to the change in Cm with the same change in alpha (and CL on the wing) is IDENTICAL.
The percentage change of the tailplane lift is utterly irrelevant to the pitch stability consideration.
Unless the tailplane is operating in the non-linear lift-curve slope range - which means it's FAR too close to stall for design comfort - there is no benefit to any particular tailplane angle.
Those people with access to a trimming taiplane aircraft could prove it if they wished. Find a flight condition where the stab to trim with no column input is in the middle of the trim range. trim the stab nose-up and apply nose-down elevator to hold the trim alpha, then do the same in the other direction - the aircraft short period response - it's pitch stability - will be the same whatever the combination of tail and elevator angles used to trim.
If tail angle affected stability, I would see different dCm/dCL gradients when I tested an aircraft in ther wind tunnel at the range of tail settings - something that is routinely done to establish tail effectiveness and downwash characteristics; in the normal operating range, where the curves are linear, there is NO effect and the curves are parallel, even with huge changes in tail angle - 15 to 20 degrees or more.