It seems to me that having analog input level on a 32 bit float recorder does risk undermining the whole point. I do feel that a key benefit of 32 bit float devices is that the whole signal chain can be optimised end to end - the mic (if built in) can be matched to the preamp, and the preamp matched to the a/d converter, using the maximum preamp gain which the a/d stage can handle, because max preamp gain gives lowest noise. On the very modest 32 bit float devices I have, the chosen level seems to be what I would have chosen myself, if that makes sense. It's almost as if the device is using auto gain control, but fixed, and set by mind reading and time travel to the correct setting at the outset...