Okay, I've looked into this. Bouyancy is broken in such a way that it leads me to beleive this has never worked in the 2.4 code fork. The bouyancy code is effectivly disabled, being conditioned on a sim global boolean which no code sets. The way things sit currently, .rdboy gets set to 0 every cycle. and .setboy effectivly does nothing. I suspect this is one of those cases where the new physics in 2.4 allowed for the potential of obsoleting this older pre 2.4 concept, but the concept of bot density was never completed and the loose ends were never cleaned up.
Fixing this is beyond the scope of 2.43.1. I would like to hear some discussion as to whether we maintain how the .setboy / .rdboy sysvars were intended to work or rather make bouyancy an implicit function of bot density which in turn would be an implicit function of mass and size which would require we give bots a way to specify their own density. I prefer the latter and have a pretty good idea about how to make this work, but 2.43.1 is imminient, so it will be a post 2.43.1 work item.
I will leave this bug open for further disscussion.