The problem I believe with your question is to which end do you define a begin and finish.
You can create iteration loops in script by using a for each var in fieldvaluelist('field1') nested in each value dimension you want to evaluate the permutations. With a where clause you can restrict the results to give you only those that match a given output
However, how do you define how the var for each value in fieldlist is incremented. In your example, originally you had 5 agents, then in your example mix you have 50, 100, 25. This does not follow a pattern.
If I had to design this I would let SPA and sales stay the same. Since only leaving SPA the same would result in an infinite number of possibilities. Then I would need to cap Agents and Days as this could also scale up infinitely. At the lowest possible level to complete, we should define that a day is the quickest, and not half a day for instance, so by default, sales / spa would give us our max agents or days (i.e. 100/2 = 50). Sales also would determine the ratio of days and agents required to deliver the same sales target at the same rate. Here's an example application of this with all permutations.
