Arnaud,
Your guess might be pretty close.
Shortly after the '35 appeared, I was part of a group of grad students at Caltech that had lunch with Bernard Oliver (VP of HP at that time), and of course the hot topic of conversation was the new calculator, which we all lusted after but couldn't afford (it cost about two months of a grad student's stipend at that time!).
One of the questions concered how the functions were chosen. Oliver said that there had been a bread-board calculator attached to a computer, and that HP engineers (of all types, I guess) were invited to create a "button" for whatever function(s) they wanted/needed, the functionality of which would then be implemented in the attached computer.
After enough folks had picked functions, their choices were basically treated as votes for the most-useful functions.
So, it may have been that they didn't know what functions would actually be included until almost production time.