Are worlds using procedural input properly?  they seem very random, not coerced like we normally do for terrain / ecosystem / biomass generation ?  for a game that leans heavily on planetary hex building (much more heavily than it's predecessors now that ship mass building is gone) it would make sense to build out planets this way:

Form a standard procedural generation for the planet abstract hex map using a few waves of noise to build a nice Fourier for height (fbm is quite good for this hit me up if you want some c# libs), then the positions and values are input into a temperature map and a moisture map, the temp and moisture map are then post processed to normalize each other and that gives you the 3 points of info you need for the surface biome map H/M/T (height, moisture, temperature), use a bicubic interpolation to get the smoothed values to build your actual planet surface tinting, and then you can model areas of a planet that should be X and Y, like maybe science is in colder areas so the science cells are more likely to be at top or bottom of a map.   The reason why this approach isn't just overproduction, it's because currently there is no cohesion at all, you end up with a zero science planet with a +3 science bonus on it, and they'll be stippled all across the planets surface which is a bit head scratchy and it makes most planets layouts worthless for adjacency bonuses, i tend to just ignore them now because if i didn't ignore them i'd not get as much of a bonus as i do by just having them stack side by side.  The chances of a planet having the exact layout that would be useful is near zero, and i love the potential of the bonuses i really want to see that come alive but doing so needs some procedural chops.

Another thing you could look into is Voronoi cells, if you'd rather set the science/money/manufactury/food etc prior to the generation of the planet then you could generate cells by the number of different types of resource you want this planet to have more than zero of then adjust the size of those cells to match the volume you want to put on the planets, then you know you have a group of "science" cells you need to make, divide the size of the planet into the number of metrics to get the cells per type and then spread them out on the planets surface with a little noise so sometimes you do need to make a sacrifice.

Don't get me wrong, this is not unusual in a universal industry sense, and it hasn't been a significant burden but having cohesive planetary surfaces would be very good

In any event 77b with Altarians is a wonderful version, albeit with my xml double hull size and no limits on attachments mod which literally keeps me from losing focus and nodding off