OK, thanks to the @andy5995's bisect we now know what's going on, and I see at least on one of my machines that undoing most of what that commit does roughly halves my loading times as well (I guess I never really noticed because I'm used for my usual huge Geany session to be slow enough that I don't attend it, and I don't restart Geany often either).
In any case, if all the reports about this slowness come from the same issue, then we're on the right tracks: having identified the issue usually accounts for at least half the work. Thanks!