Currently is_monotonic returns "don't know" if the inteval covers more than 1 cell. If the function has a big number of cells cells (64 or more in 1 dimension) this check appears too conservative, and the shading decomposition appears too tiny. An example is the test file of the bug 687713. (This problem is not a part of the problem described in 687713).
The function monotonity chack was done a long ago, but we still need a monotonity and linearity check for color conversions. Updating the bug title and passing to the owner of color spaces.
Another shading bug for Robin to look at and/or close....
Enhancement still missing in Ghostscript 9.03