I think this was simpler because `computeRoot` also operates on a few vars inside of the scope of the wrapping method (e.g. `subtreeRoots`). I think this was an optimization wrt to the recursive calls. I would merge as is but if benchmarks and readability improve or stay the same, I would not mind Evan's suggestion in a followup PR.
_Originally posted by @liamsi in https://github.com/celestiaorg/nmt/pull/260#discussion_r1648010073_