Mono,
you actually forgot 3 more moving parts.
Apart from the 3 dithering motors (one for each axis, or is it one omtor for all), there are for each axis a small servo controlled mirror to align the lasers. It moves only on alignment, if ever, and the movement is a fraction of a degree.
The IRS (the laser one) is actually a beam of light from a laser sent into a triangle bade up by mirrors and a prism. the light is deflected both ways around in the triangle, then it meets again in the prism. The small servo operated mirror corrects the light so that the two counterrotating beams are exactly syncronized when they meet again.
The syncronization is only possible because LASER light is not only only one frequency, but also in phase (like the output from an AC generator). The result is a so called "fringe pattern" which will be steady when the IRS is kept calm, but it starts moving as soon as the IRS is turned. Just like a music record player of the type with dashes painted on the disc, where a light powered by AC (a stroboscope) makes the lines appear steady when the correct speed is matched.
So, IRS still need accelerometers, as the laser part of it is only the angular displacement, not the acceleration.
Cheaper systems use fibre optics instead of the mirrors. The are a lot cheaper, but not as precise over time. I would imagine this is OK for missiles which in nature need not fly for prolonged times, and they usually need only fly once
boom
The original question as to what is INS vs IRS has many confused so both terms are used on either laser or stabilized platforms. Just like terms like FMS/FMC/CDU are all used to describe the same thing.