If I understand correctly you mean the temp differential will cause movement in the metal and problems with alignment? I think alignment will be the hardest thing to overcome in a system that is based on temp differentials.I think your diagram works well if only materials ( heathink and baseplate ) are machined by extrusion on bar block Al/Cu. If is hot laminated materials, the temperature varation have an inpact on form memory material like ) or ( if heat or cold.
Just my experience on the metal plates, and the longer it takes the more it moves.
(Thanks google to help me for translation )
I have seen papers where they have used 1 diffractive element to tune all individual diode at once. Somehow the angle of incidence has an effect on the freq. the individual diodes lock to. Then the individual beams of slightly varying wavelengths are combined via a grating. I think this may be out of range for the experimenter? The elements were made via holographic methods if I recall.Incidentally I looked at FBGs for stabilizing a seed laser at 1064, a few months ago. Best accuracy yet Affordable stuff was all +/- .3 nm. Good enough for our purposes, except 2 nm steps between red wavelengths are NOT a stock product.
I think planters and others are on to something. Andy has shown it will work. I think its a matter of finding a practical way of doing it without the esoteric gratings.