I remember correctly, the wavelength dependence of the blue diodes is much less than for red ones; for red ones it is typically 0.3nm/C while for blue ones it is like one-fifth of that. I may run a test in the next couple of days to verify this for the blue Casios.
At any rate, Tec cooling/heating a diode for wavelenght shift is hardly worthwile because you could barely see any difference (unless you cool/heat by several tens of degrees which is not advisable).
As for the title of the thread.. .. why bother if you can't decide it by the naked eye?