Clever! Any time you can minimize calculations is better, but I wonder if the user will be given the lowest grade of the two categories in every case instead of the average of the two categories once 50% is reached in both. I may have misunderstood the problem or the solution above. Either way if it works do it!
I'm not on the gradebook development team so I am not certain about this, but I think Natural is meant to be the one aggregation that rules them all. In your case (a category where there is a 5 point assignment and a 6 point assignment that add up to 100), the user would have to set the grade point values to a number (say 50) then weight the assignments to match the 5 and 6 point distribution (weights 45 and 55 respectively). You could also make those point values for the assignment 45 and 55 if you wish.
Then, if you wanted to grade out of 5 then you would use a rubric in the assignment with a max 5 grade and it would scale up to whatever number the grade is. So, in that way natural could still achieve the aggregation you mention above.
However, I don't think any of the aggregation methods are going away for some time. There are a lot of users out there that have these as parts of large enrollment online courses where messing with the grades would have a disastrous cascading effect. Again, if it's working I always say use it. If core does away with SWM in some future update, you can cross that bridge when we come to it.