Is the reason that the "offset" inlet isn't a signal inlet mainly for performance in most use-cases?
Because if it were it seems like users could give 32-bit floats to both inlets, and have them added as doubles internally?

-Seb