Presumably if you generate a table that randomizes the run order, then it would be possible for the same operator to run twice in a row. What I would do is let JMP create the random order then manually revise that run order to meet your goal. It can still be random enough.
This is acceptable if you do it infrequently. If you need to do this many times, then I understand your desire for an automated method. However, if you need this randomized run order frequently, perhaps you can develop a randomized order once then re-use it for future experiments.