I actually had the same idea earlier and finished a draft a while ago. Since the arrays have an even number of elements, you need to average two elements, possibly leading to fractional numbers.
Here's what I did. Gives about the same result as the stock median VI rounded down to the next integer in limited testing. (10M trials, never off by more than 0.5. This is just the inner median calculation, of course. I have not benchmarked it. There could be bugs). Adjustments are needed for larger arrays or odd sized inputs.
