I'm trying to use Jitter to generate a biased sine table made out of 8-bit integer values. I thought it would be easiest to start with floats and convert to integers at the end. I seem to be generating the data fine in float32 format, but I am not able to convert the resulting Jitter matrix to char or long. I have tried jit.coerce and a second jit.matrix object, but in both cases the result is garbage data. Am I misunderstanding what jit.coerce is supposed to do?