Floating point data that contains exact zeros as these require different treatment depending on whether one uses SUBTRACTIVE_DITHER_1 or SUBTRACTIVE_DITHER_2 (in fact it is the only difference between the two). Maybe our default random array can contain exact zeros by default to avoid having even more combinations (I don't think there is any harm in including that by default).
Floating point data that contains exact zeros as these require different treatment depending on whether one uses SUBTRACTIVE_DITHER_1 or SUBTRACTIVE_DITHER_2 (in fact it is the only difference between the two). Maybe our default random array can contain exact zeros by default to avoid having even more combinations (I don't think there is any harm in including that by default).