^^^^^^ What he said, with the following comment: With an interferometer, you need some degree of reflectivity, in order to get light back into the measurement system. If your parts are specular to some degree, you can measure in that manner, which does also apply to optical flat, but more so, since you are judging by eye typically, instead of a very sensitive CCD or CMOS camera. For interferometric measurement, it would probably work best with a scanning white-light profiler and very low mag objective (like 1X or 0.5X), again depending on how much light you return to the instrument. This sort of form/shape measurement range, between about 5 and 25 microns is always a problem area in my experience, where you run out of easy reliable mechanical measurement capability, and you aren't quite into the range where optics work well. And if the part is non-reflective, more complications.
My choice for doing this would be 1) sanity check on surface plate with tenth indicator, and hope for really low numbers to say it's good, 2) tenth-reading mic to check for any thkns weirdness, to corroborate test indicator measurement and exercise common sense, 3) CMM with small probe and lotsa points if 1 and 2 are inconclusive, and 4) optical profiler (in that order, if you don't already have access to an optical profiler, in which case that would be first).
