I don't know if this is useful at all. Because you/splitter don't know about dispersion of the nodes. For example take a two tiles who separate Berlin in two parts. Both tiles would have capacity of some more nodes. But if you enlarge the southern tile a little bit to north, you may get problems.
You describe an algorithm that uses the result of mkgmap processing to
"optimize" the position or size
of the tiles. Can you describe more detailed what splitter should do with
this information?
Gerd
Mkgmap can Output a List of the size of Evelyn tile. In the next Splitter Run Splitter have the Output of mkgmap an an Areas.list.So there Splitter can Take the Old Areas.list with the mkgmap Output an perhaps the input-osm-Data and Fine-tune the area-List.But this should be an option in splitter.